Skip to main content

Full text of "Architectural composition : an attempt to order and phrase ideas which hitherto have been only felt by the instinctive taste of designers"

See other formats

£ i 

m m mi \m w. 

y oMim\n^ 


%UAiNfl-3t\v ^Aavaani^ y 0Aavaan# 

l ^lOS-ANGElta* 

T O 

= <C 
-n I— ' 









1 !V©i 

y #Aava 

y 0Aavaan#- 










^/OJIWOJO^ ^toojmo-jtf* 


y 0Aavaan^ 




ry 0Aavaan# y 0Aavaan-# 






^OJIIVJJO^ %0JI1VJJ0^ <rjU3NY-S01^ ^/XMAINlrf 



^Aavaam^ y 0Aavaan-^ 




^a3AINfl3UV > 





I IJI7 1 1 1J(7 I 

32)1 1^1 Jiinfl 1MI 1 

'lDNVSOl^ "%3AlNfl-3\\v 

iki SV 

^AHvaan# y <?Aran-#- ^&h3ny-sqv^ % 



(03ITV>JO V 







AavaaiH^ 7 

y 0Aavaan-# 






^^9 1 

. W „ CO 




%JI1V3J0^ ^Sf( 

p— u 

^atoih^ y 0i 















.\WUN!VER% ^cl 

I -n <-> 

,\tfEUNIVER% ^1 





.^UIBRARY-0/* ^ 



?A«vaan^ y 0Aavaaii-^ 



^OF-CAtlFORto ^0 

y 0Aavaan-#- y o, 

^■UNIVERSjjf. vjslOSANCElfj> 



^1 liStt IJUI? I iJUir 1 li^l K 



An Attempt to Order and Phrase Ideas 

which hitherto have been only Felt by the 

Instinctive Taste of Designers 


John Beverley Robinson 





New York 




Copyright, 1908, 



! '. I V. ', 1 I]'',. 

Architecture & 
Urban Planning 




A CONVICTION in the mind of the author of the possi- 
bility of formulating the approved practice of archi- 
tects in designing the exterior of buildings resulted in the 
publication in 1898 of a series of articles upon the Principles 
of Architectural Composition in the Architectural Record, 
which were afterward reprinted in book form. 

This was favorably received, and formed the basis of a 
course of lectures which has been given by the author 
annually for some years at Columbia University before the 
students of the School of Architecture. 

The theories involved have never been impugned, nor 
indeed can they well be; as they comprise only generaliza- 
tions of principles which have long been recognized in their 
individual application. 

The present work developes these theories in more co- 
herent and logical form, and, it is believed, will serve to 
simplify the acquisition of the subject of which it treats. 

In conclusion, the author expresses his thanks to those 
whose names follow for their kind aid in obtaining illustra- 
tions: Messrs. Babb, Cook & Willard; Mr. James B. Baker; 
Mr. H. W. Desmond; Mr. Wilson Eyre; Messrs. Carrere & 
Hastings; Mr. George A. Freeman; Messrs. Howe, Hoit & 
Cutler; Mr. Elliott Lynch; Mr. Benjamin W. Morris; Messrs. 
Bruce, Price & de Silbour; Mr. George B. Post; Mr. H. W. 
Poor; Messrs. Andrew J. Robinson Co. ; and Mr. E. R. Smith. 

John Beverley Robinson. 

Architectural Department, 

Board of Education, 

New York, 1907. 

1 60Q274 



I. — The Standard of Taste 3 

II. — What is Architecture? 8 

III. — Unity 19 

\/lV. — Individuality 22 

V. — Similarity 35 

VI. — Subordination 47 

VII. — Analysis of Buildings 57 

VIII. — Primary Masses 63 

IX. — Secondary Masses 86 

X. — Details 99 

XI. — Horizontal Division 115 

XII. — Proportion 137 

XIII. — Contrast 170 

XIV. — Practical Applications 178 

XV. — Asymmetrical Composition 191 

XVI. — Flexibility of Types 202 

XVII. — Comparison and Criticism 212 



Academy of Fine Arts, Vienna. (Theophilus Hansen, Architect) 128 

Arundel Castle, Gateway . 70 

Banqueting Hall, Whitehall. (Inigo Jones, Architect) . .120 

Bourse, Lyons. (Dardel, Architect) ...... 76 

Bourse, Marseilles . 121 

Buckingham Palace, London. (William Wynne, Architect) . 77 

Chamber of Commerce, New York. (James B. Baker, Architect) . 193 

Chateau d'Azay-le-Rideau . 90 

Chateau de Chaumont 207 

Chateau de Chenonceaux ......... 206 

Chateau de Josselyn .......... 76 

Chateau de Villebon 81 

Cherbourg Apartments, New York 67 

Church of All Souls, New York. (J. Wrey Mould, Architect) . 40 
Church of St. Ambroise, Paris. (Ballu, Architect) . . . 224 
Church of St. Nicholas, Potsdam. (Friedrich Schinkel, Archi- 
tect) 148 

Schauspielhaus Theatre, Berlin. (Friedrich Schinkel, Architect) 66 

Dorilton Apartments, New York. (Janes & Leo, Architects) . 68 
Gate Lodge, North Easton, Mass. (H. H. Richardson, 

Architect) . . . . . . . . . . . .192 

Gateway, Lubeck 168 

Gateway, King's College, Cambridge 213 

Gardener's Cottage, Long Island, N. Y. (Geo. A. Freeman, 

Architect) 199 

Hotel Essex, New York. (Howard, Cauldwell & Morgan, 

Architects) 194 

Hotel de Ville, Lyons. (Antoine Desjardins, Architect) . . 65 


Liu & Deperthes, Architects) . 

. Mich. . 

. Mich. tW'ih'.n Eyre, Architect) . 
iiv, Mo. (Howe & Richardson, Architects) 
I r< ... A. Freeman, Architect) . 
Kim, M< ad & White, Architects) 
\\ V. (T. Henry Randall. Architect) . 
, Heights, NT. Y. (John Beverley Robin- 

D. C. (Carrere & Hastings, Architects) 
".; \ Coi rt House. (Frederic C. Withers, Archi 

i P. Morgan, New York. (McKim, Mead & White 


. Mass. (H. II. Richardson, Architect) 
Paris. (Claude Perrault, Architect) . 

\. Campanile 

Cathedral. (Vaudoyer & Esperandieu, Architects) 

Iame, Paris 

Building, St. Paul, Minn. (Babb, Cook & Willard 


\. Tokio 

Parliament Houses, Westminster. (Sir Charles Barry, Archi 

I. dwig Victor, Vienna. (Ferstel, Architect) . 

rEiN, Vienna 

riCE, Antwerp. (Baeckelmans, Architect) . 

ini, Florence. (Baccio d'Agnolo, Architect) 

Rome. (Antonio da San Gallo (younger) 

Florence. < P. runelleschi, Architect) . 
trozzi, Florence. (Benedetto da Majano, Architect) 

[o, Florence 

//" Vendramini, Venice. (Pietro Lombardi, Architect) 

ROME. Santa Maria dei Miracoli, Santa 
li Mont Carlo Rainaldi, Architect) . 



Porte des Cordeliers, Loches . 205 

Princeton College Dormitory. (Benjamin W. Morris, Architect) 91 

Rathhaus, Vienna. (Friedrich Schmidt, Architect) . . . 212 

Ring Theatre, Vienna 215 

School, Hester Street, New York. (C. B. J. Snyder, Architect) 150 
School, Morris Heights, New York. (C. B. J. Snyder, Archi- 
tect) 75 

St. James Building, New York. (Bruce Price, Architect) . 124 

St. Mark's Campanile, Venice 133 

Stoedel Art Gallery, Frankfort-on-the-Main. (Oscar Som- 

mer, Architect) 77 

St. Paul Building, New York. (George B. Post, Architect) . 133 

St. Paul's, London. (Christopher Wren, Architect) . -77, 160 
St. Peter's, Rome. (Michael Angelo, Architect) . . . .122 

St. Stephen's School, New York. (Elliott Lynch, Architect) . 100 

Synagogue, Berlin. (Eduard Knoblauch, Architect) . . . 149 

Taj Mahal, Agra, India .50 

Temple at Paestum 162 

Temple at Phil.e 64 

Villa Albani, Arcade, Rome. (Carlo Marchionne, Architect) . 42 

Villa Borghese. (Hans von Xanten, Architect) .... 145 

Villa Medici, Rome. (Annibale Lippi, Architect) . . . 145 

York Cathedral ,64 



THE very first obstacle that is encountered in any 
discussion of aesthetics is the old and still unsettled 
question of the standard of taste. What entitles us to say 
that anything is pretty or beautiful? 

Many things that the untutored sense applauds, the more 
sophisticated apprehension decries. The music that we now 
most enjoy, in our earlier years seemed to have no music in 
it. What we then wanted was a good old tune repeated in 
every verse ; now, both verse and tune seem repugnant and 

So in architecture. Well we can remember when our 
taste was the barbarous taste of the ordinary civilized 
human being, not able to distinguish the good from the bad, 
nor the good from the merely pretty. Looking back, we 
can now recognize that the chief source of our pleasure in a 
building was its color, the warm yellow gray of the stone in 
preference to the hackneyed red of the brick. A building 
with arches in it, too, we preferred from the earliest years. 

After all, there is an underlying accuracy in the un- 
sophisticated aesthetic sense that cannot be ignored. The 
buildings that have been admired through the ages by 
everybody appeal not less vividly to the untaught than to 
the cultivated taste. What veriest Philistine failed to 
admire the great Campanile of Venice, now so recently 
vanished from earth; yet what tower was more praised by 



i nnoisseur? Certainly where Philistine and connois- 
• e we are entitled to place our confidence. 

. too, the ever-fluctuating fashion of the day to be 

i into accnnt. For a few years we all admired Vic- 
o Gothic; then we branched off on the "Art nouveau" 
of its day, the fruitful though much-abused "Queen Anne." 
Then came Richardson and his imitators; then nothing 
would do but Italian Renaissance; until, through the road 
of the classical revival, merging into the "Colonial," we 
have brought up at the present time in the modern French 
st vie. in which, as in any other, many good and many bad 
things are done daily. 

Willi all these changes we have at least gained in catho- 
licity of taste. The time was when a sincere admirer of 
the revived Greek style really thought a Gothic church 
i< I of 1 teauty ; to-day we are at least able to put ourselves 
in other people's shoes and regulate our admiration by 
other standards than our personal preferences. 

\'o attempt is made in the following pages to lay down 
any standard of taste at all, except this: 

Any system of composition to be at all valid must be 
independent of the variations in methods that have pre- 
vailed, and must be applicable to all styles that have 
existed and to all that shall hereafter come to life. Such 
illustrations as are given are chosen intentionally from the 
most widely varying types, and it is assumed that designs 
which arc generally admired in their time and place are 
proper objects of admiration. 

It is exceedingly doubtful, however, whether our ad- 
miration of the monuments of the past is of precisely the 
should be. A Doric temple is generally regarded by 

schools as the most perfect development of architecture. 
deed, but it is very doubtful whether it is the most 


beautiful. On grounds of pure beauty cither the Japanese 
temple or the Gothic temple would probably outrank it; but 
it must be remembered that to the Greek, as to ourselves, 
"mere prettiness" was not all. Besides the criterion of 
beauty, the Greek regarded to irpeirov, the proper, or the 
suitable, as indispensable. 

The triglyphs, for instance, always seem out of place to 
the unaccustomed critic, and that they seemed so to the 
delicate Greek apprehension is evidenced by the fact that 
in the later styles, the Ionic and Corinthian, they were 
left out, and the frieze was made continuous. 

All such questions are left untouched. It is assumed 
that if the triglyphs w r ere good enough for the Greeks to 
use, they are good enough for us to regard them as a proper 
adjunct of a carefully considered composition. 

There are two ways of presenting a subject in writing: 
the first — the analytical way — by collecting all the facts, 
laying them before the reader and suggesting the conclusions 
to which they point. The other, the synthetical way, is to 
place the facts as illustrations, after stating the general 
principle that has been reached by analysis. 

The first way is the true scientific method; its only 
drawback is that it is intolerably tedious. Neither is it in 
accordance with the natural working of the mind. Thought, 
to exist at all, must classify ideas as they come. Some 
basis of classification must previously be found, however 
temporary and incomplete, before thought can work at all. 

The second method therefore is the one that has been 
adopted. Generalizations that have been reached slowly 
by the multiplication of examples are first stated as elemen- 
tary principles and then illustrated by examples more or less 

This method involves a certain arbitrary tone in state- 


ments, which is unavoidable, yet the farthest possible from 

ittitude which is intended. With every seemingly Laid 

there is an implied appeal to the knowledge of the 

which exists in the reader beforehand, and the 

implied question : *' It" you do not think the facts point to this 

realization, to what other generalization do you think 

they point ?" 

I- • ict, it is primarily to the professional designer in an 
or to the would-be designer in office or school, that 

this book is addressed. 

ry draughtsman who develops into a designer 
advances by absorbing certain methods, scarcely conscious 
that he has done so; using one motive for all of his designs 
until another is added to his repertory, and so on, gaining a 
critical sense with each step of growth, but growing as the 
plant grows, without introspection or reflection, without 
any intellectual realization of the processes of his aesthetic 

Such a measure of skill as intuitive apprehension can 
give is properly called an art ; with reflection and generaliza- 
• .n added it becomes a science, more or less perfect. If 
any designer will join with his condemnation or approval of 
any composition the question, Why is this pleasing, or, Why 
is that unpleasing, and make, in reply, some general state- 
ment of his reasons, he may arrive at all the conclusions 
in reached, though he will very possibly use other words 

lescribe thernj 

of this work is to excite thought, and to 
register the results of the author's own observations, with 
the view of raising the power to design from its compara- 
tively obscure position as an art to the dignity of a science. 

Hitherto the designers of engineering works have 
ted that they were guided by pure science, and have 


scorned even an attempt to make their productions pleasing 
to the eye; but for the future it will be possible for the 
student of engineering to give the same sort of intellectual 
consideration to the appearance of his work that he now gives 
to its stability. 

One more reservation must be borne in mind. Among 
the various qualities that are needed for a perfect result in 
architecture it is rarely possible that all can be completely 
reached in a single design. 

The requirements and conditions of use are so exacting 
that after satisfying one aesthetic condition we are often 
obliged to dismiss others with scarcely an attempt to meet 
them, with full realization of the solecisms that we are 
forced to commit, although a skilful designer often can 
make the presence of one quality atone for the absence of 
others, concealing unavoidable disproportion of certain 
parts by accentuation of others, or modifying an unmanage- 
able plan to the advantage of its convenience in use, as well 
as of its availability as the basis of his composition. 



IN the popular mind, outside of the larger cities and the 
more cultivated circles, there is but little difference 
in architect and a builder. The writer, who deems 
himself worthy to be ranked among the former, has more 
than once been introduced to rustic committees as "the 
liter," ni >r is a much higher ideal prevalent even among 
the more sophisticated. 

At the best the architect is conceived as a builder who 
lastered the mysteries of " style." That the mechanical 
adaptation and construction of the building should be perfect 
and the "style" an accurate reproduction of the details of 
smne historical or contemporary type is the highest possible 

Even among architects themselves there is scarcely any 

broader conception. "Architecture is decorated construc- 

ion"; "Architecture is dependent on construction and 

ould express the construction"; "Architecture is suitable 

arrangements to meet certain wants." 

Planning, materials, construction may all be studied and 

alized, but, when it comes to the real architecture, the 

composition of the exterior, the only rule at present is, " Go 

and find something that is the fashion and make your 

Xow all these ideas and precepts have a 

truth. Architecture certainly implies building, and 

buildings built is a necessary condition, an 

is t( i that extent a builder. Yet that there is more 



to architecture than mere building, we all feel intuitively; 
perhaps he comes nearest to it who says : " I know what kind 
of a building I want myself, but I must have an architect to 
make the outside of it." 

In the same way architecture implies construction, and it 
is impossible to divorce the architectural result from the 
construction entirely even if it were desirable ; but that the 
exhibition of the construction is a necessary characteristic 
of good architecture cannot be maintained. 

Thus if we are obliged, or prefer, to span an opening with 
a lintel, it is sometimes permissible, or even essential, to 
cut the lintel into imitation voussoirs of a flat arch. It is 
true that the best designers try to avoid such expedients, 
but avoidance is preferable — not because deception is morally 
wrong, for deception to give pleasure is not felt to be morally 
wrong, as in the case of the beer mugs filled with jelly and 
crowned with meringue instead of foam, which are sold in 
the streets — but because of a certain sense of unreality, of 
mere theatrical pretense, which detracts from the full en- 
joyment of a design in which such pretenses are too frequent 
or too evident, however beautiful it may otherwise appear. 

Nor, again, will the frequent theory that decoration 
must not be constructed always hold good, as we admit 
when we find the culminating glories of the two great 
historical styles, the colonnade of the Greek, the spire of 
the Gothic, both of them constructed decoration, not to 
speak of many of the minor parts that were no less so. 

Neither is the material of construction a fundamental 
consideration, except in so far as it forces its consideration 
upon us. If it were possible, for instance, to carve a marble 
figure with the attenuated parts that it is possible to use in 
bronze there would be no objection to doing so, nor would 
there be any artistic objection to executing an iron grille 


■ thai the practical error would soon correct 
rapid destruction of the frail wooden frame. 
nversely, is there any objection to imitating a 
rble figure in bronze, which is perfectly practicable, saw 

■ thereby we Lose both the special beauties of which 

pable, and at the same time the special charm 

;tone in its own appropriate place. 

To gain a sufficient notion of the proper extent to which 

a material must be considered, observe and talk with a 

mi of taste engaged in designing a costume from a 

■ >n plate. N"< >te I hat the extraordinary contorted fashion 
plates are to the feminine mind merely diagrammatic, not 
intended to really represent a clothed human figure, but 
only so much stuff, with so many yards of trimming, cut to 
attain a certain result, which is always accentuated by 

ration. This costume, she will explain, must be 

uted in a material somewhat like that shown. A solid 

color would not do without more trimming; this design is 

made plain because it is of richly figured brocade. Besides 

is, it must be of a certain weight; a light summer silk, 

ver suitable in color and figured ornament, could never 

interpret those heavy folds pro] xtI v. 

S< i in architecture, there is no valid reason why a building 
uted in cement should not deserve our praise as much 
; it it were of Parian marble, except that delicacy of arris 
tlie brilliancy that results from undercutting are un- 
able in cement, and on the other hand certain charac- 
I s may better be obtained in it by forsaking the 
ition of stone and adopting a new treatment suited to 
culiar finalities. 

from such necessary considerations, an archi- 
ral object is to be regulated by the same rules and 
•ised by tin- same formulas, whether built of granite or 


stucco or sugar paste. It is not worth while to combat more 
seriously such incidental truths and partial statements of 
facts that have been allowed to obscure an intelligent theory 
of the scope and full meaning of the word architecture. 

It is universally admitted that architecture is a fine art, 
that is to say, like other fine arts, something more than a 
mechanical art. To say that sculpture was the art of mixing 
clay and shaping it would be quite inadequate. So it would 
be inadequate to define painting as the art of applying pig- 
ments to surfaces. The first might suffice to define a brick- 
maker or a potter, the second to define a house painter, for 
these are both mechanical arts, but a fine art is something 
more than this : it is a mechanical art with the added quali- 
fication that the objects produced by it must give pleasure. 

This it is that constitutes an art a fine art : its pleasure - 
giving quality independently of its useful qualities. A 
mechanical art is one which fashions objects primarily 
intended for use ; a fine art that which is primarily intended 
to give pleasure. 

Of the recognized fine arts there are three that are 
representative, either entirely or chiefly : literature, painting, 
and sculpture. These give pleasure chiefly by the pleasurable 
representations they depict, although the pleasure that 
attaches to the material form is also essential to each in 
differing degrees. Thus prose literature and poetry give 
pleasure partly by that which they describe, partly by the 
agreeable arrangement of sounds, partly by the sentiment 
that accompanies them. 

In the same way painting and sculpture produce pleasure 
by appealing to various senses and sentiments. These are 
therefore to be classified by themselves as representative 
arts in contradistinction to those remaining and now to be 


3 distinguished as pure, as opposed torepre- 

• cture and music. The office of the 

nge sounds that shall be agreeable to the 

the former is to shape material objects so that 

". pl< ase the eye. 

true that certain sentiments are aroused by certain 

h of music and of architecture: we speak of both 

: or dignified or playful or exuberant; nevertheless, 

•m the question, and a very curious one it is, of why 

ain - produce certain emotions, it is indisputable 

only means wherewith music can produce its result 

>und — absolutely nothing else . 

Iso architecture, apart from the figures, scenes and 

ornaments that may be placed upon objects 

'. by it, has its own function in fashioning material 

- into agreeable shapes, not incompatible with, but not 

rily limited by, their useful applications. 

Lusic is the art of sound, so architecture is the art of 

form. Not representative form, not garlands and metopes 

and inhabited niches, but walls and roofs and columns, and 

not only these, but objects not properly buildings at all — ■ 

Is and tripods and monuments. 

1 ely connected with architecture, forming, in fact, 

a part of it on a smaller scale, is the fictile art which 

vases, urns, and such minor objects, which may 

be put to some use, but which are intended primarily 

• looked at simply for the pleasure that their shape or 

- both combined, give. 

Another pure art, also classed as a fine art, though not 

ual in dignity to architecture and music (it has 

been ranked alongside of music), is the art of 

ble motion, the dance. 

•uld be indeed a fourth pure art, the art of non- 



representative color, bearing the same relation to painting 
that architecture does to sculpture, but this has not yet been 
separately catalogued and is usually merged in the sister 
art of form. 

This distinction between pure art and representative art 
is the reason why it is possible to lay down rules for the 
former, but not for the latter, as these are too complicated 
in their appeals to the emotions for any but the most ap- 
proximate rules, derived from those for the corresponding 
pure arts, and applicable to representative art only in so 
far as the latter must include a certain proportion of pure 
art. So the catalogue of fine arts will stand thus: 



The Art of Form ..... Architecture. 

Comprising all fictile art that is not rep- 
resentative of life, and including the 
art of decorative color. 

2. The Art of Sound Music. 

3. The Art of Motion Dancing. 


i. The Art of Representative Form . . Sculpture. 

2. The Art of Representative Color . . Painting. 

3. The Art of Representative Sound . . Literature. 

Including both prose and poetry, of 
which the former is almost entirely 
representative, laying little stress on 
the beauty of expression, while in the 
latter the importance of the expression 
is paramount. 

These are, of course, involved and intertwined in the 
most complicated manner, each one infringing on the 



of the other in a way which suggests an intimate 
connection anion- them. 

Thus Song is a combination <>f Music and Literature; 
nd the Dance arc inseparable; while Architecture 
s with itself the applied arts, so to speak, of Sculp- 
ture and Painting. 

Nevertheless, the classification which we have made is a 
nal distinction and useful in clarifying our ideas upon the 


We must therefore think of true architecture, not as the 
development of economical planning, not as the expression 
of construction, not*as adherence to historic or contemporary 
-Knt. hut as the fundamental art of inventing and 
constructing objects that please by their intrinsic form and 
a »1< ir, addressing itself to buildings in the largest sense of the 
word, whether inhabited or built only to be looked at, as 
triumphal arches, mausoleums, domes, towers, and spires. 

It is tlie combination of the requirements that usually 
architectural objects must be both habitable and also 
pleasing to look at that causes many of the difficulties of 
architecture in practice. 

The c< msiderations that we have advanced will sufficiently 
justify the use of conventional ornament in architecture 
anil refute the position so strongly taken by Ruskin, that 
the only proper ornament is that which represents some 
natural objeel . 

S.i far from using natural forms for architectural decora- 
tion, the Greeks, who are still our masters in such things, 
achieved their greatest successes in their studies of pure 
form in ornament. Their wonderful anthemion device, 
their still unsurpassed egg-and-dart invention — I say in- 

ition advisedly, in spite of its undoubted historical de- 
velopment from the lotus, as shown by Goodyear's investiga- 


tions — their equally wonderful Ionic and Corinthian capitals 
— all of these are not, and are not meant to be, either repre- 
sentative or even suggestive of vegetable form, any more 
than the profile of one of their vases is derived from or meant 
to suggest a crook-necked squash. On the contrary they are 
studies of beautiful form in the abstract, as all architectural 
objects, large or small, must be, limited only by the un- 
avoidable limitations of use and material. 

The only art that stands alongside of architecture as a 
pure art, is music, and that there is an intimate connection 
between them has long been intuitively felt. It appears 
that the Greeks associated them closely, and it is probable 
that the connection lay in some mathematical method of 
determining the form of the parts of buildings derived 
from the known numerical relations of musical intervals, 
which caused the Greeks also to associate music and mathe- 

Just what these rules were we do not know, but it seems 
certain that they must have been very flexible, and very far 
from cut-and-dried, to judge from the liberties the Greeks 
took and the continual variations they employed even in 
fixed types. 

What concerns us immediately is that, of the pure arts, 
music has for a long time been reduced to the more or less 
scientific rules of counterpoint and harmony. These rules 
when first outlined were far from scientific, being based upon 
the arbitrary distinction of chords and discords, although, 
from the first, flavored with a scientific tinge from the 
known mathematical relations of the lengths of concordant 
and discordant strings. 

Afterwards, when the length of the string was shown to be 
a function of the number of vibrations, and when the co- 
incidence of the vibrations was shown to be the funda- 


of harmonious combinations, music was placed 
•i a thoroughly scientific basis. 

rtheless, scientific though music has become, and 
s is a knowledge of the science of music to the 
r, it is far from possible to write a musical composi- 
tion entirely by rule; on the contrary, the composer writes 
by his feelings, with hardly a thought of the science of it, 
partly using his scientific knowledge unconsciously as a 
; partly referring to it in ease of doubt; partly boldly 
transgressing it, when genius leads the way. 

Observe, however, that a comparatively small amount 
needed to regulate and illumine; that is to say, 
not a small knowledge of all known musical science, but 
that all known musical science is capable of accomplishing 
very little in manufacturing ready-made musical compo- 

It is still quite impossible to sit down and turn out com- 
positions of any length and of all styles by rule; the musical 
impulse must first exist, and drive to the expression of it, 
an expression which is only facilitated by musical science. 
Is it not possible for us to lay down such a measure of 
nee for the other pure art, architecture? Is it not pos- 
sible for us to select certain combinations that are known to 
be pleasing and distinguish in them the simple facts of 
numl >er and size that mark them, as the first musical theorizer 
n-d the ] 'leasing chords and measured the strings that 
sang them? 

It is not to be supposed in making this comparison with 

music, as that one of the pure arts that first reached scientific 

li ipment, that any strict deduction from one pure art to 

the other is intended or is possible; yet as the eye and the ear 

he only two senses to which measured a:\sthetic pleasure 

sible, tlie one in feet and inches, the other in bars and 


SBea-d. and TSeel- 


beats, it may be that the rules which apply to one may 
eventually be found to be based upon the same psychological 
facts as those which apply to the other, and the ground may 
be laid for a general theory of aesthetics which will include 
all the arts of representation, as these are but the pure arts 
of sound and form, with an element of portraiture added. 

Note some interesting analogies which may mean much 
in future. 

Why is it that in all the arts a unit composed of one 
large and two small parts gives pleasure? 

In architecture, the bead and reel in all its varieties. 

In poetry, the typical dactylic measure, the mere 
diagram of w h i c h 
looks almost like a 
bead and reel orna- 

In music such 
passages as this, 
from Rubinstein's 5 v H*^ "^ ^ 

Melody in F. 

And in the dance, the ever favorite deux temps, or two- 
step, as it is now called. 

Is it possible that such curious likenesses are destitute of 
meaning ; or is it not probable that they point to an under- 
lying unity of the aesthetic arts that we shall some day 
penetrate ? 

We are about to develop certain theories which may 
serve as a foundation for architecture, very much as the 
theories of harmony serve as a foundation for music. 

Not that such theories will enable anyone to conceive 
and elaborate an architectural creation any more than the 
scientific theories of harmony can originate musical com- 
positions ; the best that we can do, whether in form or sound 


ZDuctylic Weaswe 

IMelvdy in 2^ 


., is to lay down such rules as will facilitate the 
material expression of the conceptions of the imagination 

and reason : 

That done, architecture will take its true place as an art 
founded upon science, no1 only in the subordinate functions 
of arrangement and construction, but in the fundamental 
function, l>v which alone architecture can claim rank as a 
fine art at all, that of making its creations beautiful. 



IN all works of fine art there is one fundamental quality 
which from antiquity has been recognized as essential. 
This quality is unity. 

In literature this means sticking to the subject in hand, 
and not running off into excursions on allied topics ; besides 
this there is unity of literary form, to which in poetry the 
metre so much conduces. 

In music, unity lies in the persistence of a single theme 
throughout a composition, to which add the formal unity of 
key and measure. 

In painting, apart from the unity of arrangement which 
is closely analogous to that of architecture, there is the unity 
of color that comes from painting every part with the same 

In the arts of design, including architecture, one of the 
chief sources of unity lies in the arrangement of parts, by 
which objects otherwise unrelated are so placed that the 
mind loses sight of them as separate objects, and notes only 
the combination as a single whole. 

Thus a number of lines taken at random and laid in no 
particular order, cannot impress the mind otherwise than as 
a multitude of objects {a, Fig. i). 

Placed thus (6), radiating from a centre, the mind regards 
the combination as a single star or flower, and forgets to 
enumerate its parts at all. 

So such forms as these (c) remain isolated individuals 



combined in a honeysuckle (</). Another 
' unity is the intrinsic power of certain forms when 

properly placed. Thus, 
\\J// in an enriched moulding 

'/ \ / \ ~^7/\^F khe forms of the enrich- 

— W \ ' ment acquire unity mere- 

ly by their arrangement 
in a straight line (e), just 
as in (b) unity was given 
by arrangement in a cir- 
cle. The addition of hori- 
zontal straight lines on 
each side, as at (/), at 



Unity produced bya^rlngement of parts. once g iveS a complete 

union of the parts, so 

that an observer, if asked to describe what he saw in (/) 
would answer, a border, or an ornamental band, and not, 
ten ovals, eleven dashes, and two lines. This power of a 
straight line to give unity is constantly used in architecture 
in the horizontal mouldings, whether enriched or not, 
which arc frequent in all styles. 

So fundamental is this quality of unity that scarcely any 
other is needed for excellence in a work of art. If we can 
in producing something in which no part seems to 
e extraneous "stuck on," "gingerbread" — such are the 
colloquial criticisms— we may be confident that we have 
done a fairly good thing. If, in addition, every part is ex- 
actly suited t< • its place in shape, size, and relative dimensions 
suit will be not far from perfection. 

[ualities as conduce to the beauty of the com- 
position, such as grace, delicacy, refinement, are too elusive 
labeled talogued; the only criterion that we shall 

eal to in the following pages is this quality of unity. 


In every architectural object, whether great or small, 
simple or complex, whole or part, from the pyramid to the 
denticule, there are five aspects of its material shape that 
must be separately considered as related to the unity of the 

First, intrinsic shape ; and we shall find that certain shapes 
must be used in certain ways to produce the effect of a single 

Second, relative shape, that is to say, the shape of each 
part in relation to other parts, whether of similarity or of 

Third, relative size, one or more parts predominating 
and the rest subordinated by their inferior size. 

Fourth, number of parts, with their size, shape, and 
relative position taken into consideration. 

Fifth, relative dimensions of parts, or what is properly 
called proportion. 

In the following chapters we shall take up these subjects 
and deal with them more fully in the order in which they 
are here enumerated. 



I\ ; ,ii architectural compositions there are two sorts of 
shapes which correspond to two opposite sentiments. 
\. >i i >nly are there two such sets of shapes, but there are also 
corresponding arrangements of parts in respect to number 
and size which arouse answering emotions. 

These emotions or sentiments, using the words as ex- 
pressing the same feelings in different degrees, are the senti- 
ment which we have previously dwelt upon, that of unity, 
and its antithesis, by which is meant not heterogeneity or 
disorder, but rather the opposite of that strong sense of 
fixity upon a single object which constitutes unity. Shall 
we say then a sense of plurality? It is hardly that in all 

3, as plurality relates to number only, while we shall 
find that a sense of individuality and its opposite attaches to 
shape alone as well as to number. 

Indeed there is no more basic mind-principle than this, 
or than these, for they are counter faces of the same mental 
make-up that runs through all architectural, and indeed all 
Other kinds of design. 

'Ida's antithetical view might be called infinity, as op- 
d to individuality, but that would hardly give the true 
frding. What is to be expressed is the sense of spreading 
ipposed to concentration. Carried to its limit this 
spreading out does become infinity, but it may exist within 
narrower limits, and for such more frequent use needs a 
more descriptive name. 



Let us call it then distribution or continuity, distribution 
more especially as applied to the number of objects, when 
we come to speak of that aspect of them, while continuity 
may serve better with respect to their shape, with which we 
are now concerned. 

Yet we may use these words, continuity and distribution, 
sometimes almost interchangeably, the ideas which they are 
to convey being closely locked, as merely different parts of 
one and the same sentiment, and always tending toward the 
ultimate emotion of infinity. 

Individuality, while closely connected with unity, is not 
quite identical with it. Unity may exist in a composi- 
tion of which some parts are marked by individuality 
and others by continuity ; or continuity may be the most 
striking characteristic of a composition, which, as a whole, 
possesses perfect unity. Where the composition is simple 
and the parts few, individuality becomes identified with 

The Egyptian pyramid, rising from the level ground, 
and with every outline trending toward a single vertex, 
with pyramidal and pointed forms generally have been rec- 
ognized as possessing the most striking individuality. 

Everywhere in literature have pointed forms figured 
even as sentient beings in metaphor, so strongly is this feel- 
ing of individuality inherent in their shape. 

Hast thou no voice, O Peak? 

It has not been commonly recognized, but reflection will 
show that it is true that the only other architectural termina- 
tion of a distinct character, the horizontal or square-headed 
type, is absolutely without individuality, and is the fullest 
expression of its antithesis, namely, continuity. In looking 
at an object with a horizontal line above, the eye ranges from 


le, m »where finding a point of concentration, and in 

nee of concentration there is no individuality. 

in architectural composition is it necessary to say 

'; in general composition a horizontal line, or indeed 

is line of any kind, has this same character. 

in composition the bottom line is always 

-mined by the ground the building stands upon, so that 

initiation above is all that we have to consider. 

Thus our first classification of architectural objects into 

pyramidal and square-headed, which seems at first glance 

superficial and trivial, is really basic. This will be seen more 

clearly if we trace its origin and consequences a little further. 

The reason why a pyramid possesses individuality is 

use the lines trend to a single point. Now if there is 

any other way of indicating a point we may find a clue to 

other expressions of individuality. In general composition, 

that is to say, decorative design on the flat, that knows 

neither up nor down, the most individual object is a point or 

a circle. But in architecture, until we come to details, such 

as panels, windows, carvings, and the like, which may be 

circular, we have not the material conditions for drawing 

circles and points. All of our architectural objects must 

tand on the ground, and our boundary lines on each side 

must be approximately vertical, as become stones and 

bricks that are piled on top of each other. 

In tlie plan, it is true, we may have circles, and a certain 

ttaches to these even when they are not seen, but 

r inferred, as they always must be. The view of a 

Jar temple at a moderate distance could be perfectly 

! by a very slight convex curve in the entablature, 

e plan remaining a straight line, and by setting the columns 

toward the ends of the colonnade, all of them 

ling in a straight line. 



It might be necessary to render two or three terminal 
columns, where they run together, by single ones appro- 
priately cut, but if properly done, and lighted full in front to 
minimize the effect of shading, a circular temple could be 
depicted with a flat-faced structure of stone just as well as 
on a flat canvas with paint and brush. 

For any true aesthetic judgment of the productions of 
architecture we must judge them as we do a picture, ignoring 
considerations of utility in the plan and practicability in the 

Now, from this point of view, if a building with a hori- 
zontal top line lacks individuality, and possesses continuity, 






Fig. 2. 
Unity produced by either individuality or continuity. 

as at Fig. 2 (a), tending more and more to infinity as at (b), 
the longer it is in proportion to its height, what will happen 
if, instead of lengthening it out to infinity, we shorten it 
until it becomes very narrow relatively to its height? 
Evidently as we narrow the base we attain more and more 


individuality, until a1 the Logical limit of a perpendicular 
line we would reach the next thing to a point, in fact, the 
possible approximation thereto. 

We thus note that the architectural object next in 
individuality to the pyramid and pointed shapes generally 
is the tower. 

Now this connection of the long horizontal building with 
the idea of continuity, and of the tower with individuality, 
leads to a further generalization. 

Every architectural structure is naturally composed in 
its vital structure of horizontal and vertical lines. Vertical, 

iuse the fi tree < if gravity requires things to be piled plumb 
on top of one another in order to stand at all; horizontal, 
because the level ground and floors, aided by the beds of the 
construction, whether of wood, stone, or iron, naturally so 
determine ; so that in every structure there are both of these 
principles, that of individuality, centring in the perpendicu- 
lar line and perpendicular mass, and that of continuity, 
finding its expression in the horizontal line and protracted 
horizontal mass. 

It must not be supposed that because continuity is 
antithetical to individuality it is therefore antagonistic to 
unity. On the contrary, quite the opposite is true: con- 
tinuity of line and mass, when properly used, is quite as 
requisite as individuality for unity, indeed may often take 
the place of it entirely, as in any colonnade, for instance, 
lat lacks a pediment. The introduction of the pediment 
introduces the element of individuality in so far as it brings 

i pyramidal outline, but without the pediment there is 

n< .thing < >f what we have called individuality in the colonnade ; 

s unity is obtained by the horizontal lines of stylobate and 

superstructure, as well as by the horizontal succession of 


Most essential to the unity of a building are these long, 
uninterrupted horizontal lines. The base of a building is 
structurally designed to support a certain weight, and 
intellectually to appear capable of supporting it, but from 
a purely assthetic standpoint its function is not that of a 
vertical support, but of a horizontal tie, as it is to strike the 
eye, irrespective of the mind, as a living something which 
holds the columns together and prevents them from march- 
ing away out of ranks, rather than as a mere block for them 
to stand on. 

And this is true even more forcibly of the main cornice 
and the intermediate cornices, or string courses, or balconies, 
or whatever the projections may be that make long horizontal 
shadows. They are all aesthetically ties that bind ; their 
structural function is of no importance, except this, that 
afterwards, when the mind has time to look into it and form 
an opinion concerning it, there must be nothing which shocks 
the judgment, as would a cornice of paper, or a string course 
carefully painted and shaded. 

But structural considerations have no weight in the first 
conception of an architectural composition. It is true that 
the experienced architect thinks constructively, so to speak, 
and modifies even his dreams unconsciously so that they are 
capable of being built ; but he dreams in pictures, rendered 
in black and white and tinted with grays and browns and 

A cornice is structurally the edge of a roof; pictorially 
it is a broad, black line. A string course is of little or no use 
structurally, but is introduced solely for the sake of the 
minor black line that is drawn by means of it. 

In the same way a pilaster, or engaged column, is 
structurally intended to support concentrated weight, and 
must appeal to the judgment as suited to the purpose. 


irially, it is but a modeled, vertieal line, built for the 
•f the shad< >w it easts, and serving to cut the facade into 
standing side by side, which again are bound together 
he horizontal lines. 

Although we ignore the structural point of view in this 
book, it is only as a treatise on construction must ignore 
i i ic side : the two aspects of the subject, indissolubly 
united in practice, must be studied independently. 

Several practical precepts result from what has been 
lid. In the first place, to give individuality to a building, 
a i tart < >f a 1 uiilding, the pyramidal outline is peculiarly 
suited. Whether of Egyptian proportions or elongated into 
thic spire, or flattened into a gable, or depressed into a 
mere pediment, the pointed outline marks the part to which 
t is applied with an especially forcible sense of concentra- 
•• individuality. 
The same is i rue of domes, lanterns, and all approxima- 
t i< nis to the pyramidal. A number of buildings rising gradu- 
ally one above the other, 
about one highest central 
building, like those on the 
fortified rock of Mont Saint 
Michel (Fig. 3), are united 
by this disposition into a 
single mind-picture. Even 
where the situation is not so 
striking, any ordinary group 
of heterogeneous parts can 
be " pulled together" if one 
of them can be arranged as 
ver around which the rest cluster, as in Fig. 4. 
1 1 for any reason the pointed outline is unavailable and 
im] lelled to use the square-headed outline, it is quite 



Fig- 3- 
Mont Saint Michel. 

Effect of pyramidal arrangement in 
giving unity. 



possible to give individuality to a square-headed mass by 
proper treatment; only we must remember that the square 
outline has not the character of individuality in itself as the 
pointed outline has, but 
requires special treatment 
to give it that character. 

The second precept is in 
connection with the rela- 
tive height and width of 
the building. 

Buildings of any di- 
mensions may have either 
an individual or a continu- 
ous treatment ; but those in 
which the height is greater 
than the width lend them- 
selves to the individual 
or vertical treatment with 
greater facility, while 
those in which the length 
surpasses the height are 
naturally suited to a con- 
tinuous or horizontal treat- 

Of course in all buildings, whatever their relative dimen- 
sions of height to length, both horizontal and vertical lines 
must occur, but it is for the designer to lay stress upon one 
set or the other of these lines. Hence the second precept 
is to determine in advance whether the conditions require 
a relatively high building or a relatively wide building, and 
adapt the treatment thereto, using strongly marked hori- 
zontal lines, and vertical lines either not at all or very much 
more lightly drawn for the wide and low building, and, vice 

Fig. 4. 
Jefferson Market Court House. 

Unity given by a single dominating 


strong vertical and lower and lighter horizontal for 

• high and narrow building. It is possible, indeed, to 

the treatment, and to design a tower with heavy 

ontal lines, but more time and skill are required to 

obtain satisfactory results. 

Beware, though, of mixing the motives of vertically 
and horizontality. 

[f we are to use a heavy projection of cornice, giving a 
marked horizontal treatment to our building, it is most 
difficult to introduce a vertical part, as a tower, in connec- 
tion therewith ; the unfortu- 
\ nate result of such a combi- 

BJro nation is shown in Fig. 5. 

." fjj«t For this reason in the Gothic 

styles, where vertical lines 
and masses predominate, 
the heavy cornices of the 
classic orders are shrunken 
into mere stringcourses. 

On the other hand the 
difficulty of assimilating a 
tower with classic treatment 

Injudicious combination of vertical • 111 „ -t 4. u ^^ 

and horizontal motives. « well known ; if we set it on 

top of the building the judg- 

ment is < iffended ; if we carry its lines down to the ground, as 

the judgment demands, the strong verticality is antagonistic 

to the powerful horizontal lines of the portico and cornice. 

It is true that by skillful treatment much may be done 

to unify a combination of a vertical mass and a horizontal 

Either one may be characterized by predominating 

horizontal lines or by predominating vertical lines ; and if the 

same treatment is used in both the effect of unity in the 

composition may be obtained. 




Thus in Fig. 6 we have a combination of a horizontal and 
a vertical mass with three different kinds of treatment. In 




Fig. 6. 
Methods of harmonizing vertical and horizontal treatment. 

a, both the horizontal and the vertical masses receive 
square-headed terminations marked in both cases by heavy 
machicolated cornices. 

In b, both receive pointed terminations, the projection 
of the mouldings whence these originate being kept very 
slight. In c, the whole is cut into vertical slices by vertical 
lines, "pulling the parts together" effectively and com- 
pletely. Although these are shown as pinnacled projections, 
any other sort of vertical lines, as pilasters or colonnettes of 
any description, would give a like result more or less com- 

Should an individual or vertical object be attached to one 
of strongly marked horizontal character the result is often 
most unfortunate, yet such a combination as that shown in 
Fig. 7, a, is frequently perpetrated by the builder who is his 
own architect, although no architect worthy of the name 
could be guilty of such an error. It is extremely interesting, 
however, when some such atrocity committed in the name 
of architecture is encountered, not to rest content with 
denunciation, but to question and analyze it forthwith. 


s this thing so ugly? What general principle does it 
trai 1 low can it be improved? 

a very profitable exercise to assume the same con- 
as, in the above instance, an oblong block of a 
building with a turret on the angle, and observe what sort of 
treatment becomes necessary to make it presentable. It 
will be found that the difficulty lies in the reconciliation of 
the pointed termination with the strong horizontal cornice. 
If the pointed termination of the turret is removed and a 
corresponding cornice placed upon it, the first step toward a 


n n n n n o 



n B P9PI 

a b . c 

Fig. 7. 

a and c. Improper combinations of vertical and horizontal motives. 
b. Method of harmonizing such a combination. 

solution is taken (Fig. 7, b). Then, by stopping the heavy 
horizontal line where the turret occurs, a reasonably satis- 
ry result is secured. 

A similar error is seen in Fig. J,c, and it is one which, with 
various modifications, often occurs upon street corners. 
It usually results from the desire on the part of the owner to 
advertise the business by a striking architectural feature, 
combined with a realization on the part of the designer of 
e f( 'Tee ( »f the constructive dictum that the lines of a tower 
should spring from the ground. 

As in the previous example, the strong horizontal mass 
of the building is absolutely .antagonistic to the equally 



strong vertical mass of the turret, accented in its individual- 
ity as it is by the dome-shaped termination. 

The difficulty of combining pointed forms with horizontal 
ones is often felt even in monumental buildings of academic 
design, where pedimented 
forms are introduced 
above the main cornice, 
as in Fig. 8, a. It be- 
comes almost imperative 
to carry through the hori- 
zontal line above these 
pediments as at b. Such 
forms are wisely avoided 
and square terminations 
used instead. 

To recapitulate : First. 
As architecture is, in con- 
struction, fundamentally 
the placing of horizontal things to be supported upon verti- 
cal supports, even the arch being but a way of solving the 
problem that includes both factors, but does not eliminate 
either, so the fundamental aesthetic conception in archi- 
tecture is this essential distinction between vertical and 
horizontal motives. 

Second. All objects in which the vertical dimensions 
surpass the horizontal are most easily and naturally adapted 
to a vertical or individual treatment, with the limit in the 
direction of verticality marked by the solitary tower with 
pointed termination. On the other hand, as soon as the 
horizontal dimensions exceed the vertical, we may naturally 
turn to a horizontal treatment, with strongly marked 
horizontal lines and a square flat top, the extreme case 
being that of the classical colonnade. 

Fig. 8. 

Horizontal terminal line as at b, 
preferable to accentuation of vertical 
parts as at a. 


In many cases, however, it becomes necessary to combine 
a vertical form, as a lower, with a horizontal mass; con- 
which demand similar treatment for both the 
horizontal and vertical parts. 
Third. Even in single 
masses, whatever their rela- 
tive dimensions of height to 
length, either individual or 
continuous treatment may- 
be used, according to the re- 
sult we may wish to attain. 
Jf the object be to give strik- 
ing individuality to the build- 
ing a single gable may be used, as in a, Fig. 9, at the expense 
of the apparent horizontal dimensions. But if it be desired 
to make the most of the horizontal dimensions the opposite 
• ment, as at b, must be adopted, and everything done to 
diminish the apparent height and to reinforce the continuous 
horizontal lines. 

Fig. 9. 

Opposite modes of treatment. 
.; Vertical and individual, b. Hori- 
,1 and continuous. 



THE second principle from which unity of the whole 
composition springs is similarity of the parts of 
which it is composed. We are speaking now of similarity 
of shape, not mathematical similarity, still less as implying 
anything like equality in size. 

Similarity of shape should prevail among all the parts 
of a composition, 

from greatest to > * > ^~s> 

smallest, from dome 
to door panel. 

To illustrate 
what similarity 
means, let us see 
what dissimilarity 

Such composi- 
tions as that shown 
in Fig. 10, a, are by 
no means unusual. 
After recessing the 
front, perhaps to 

obtain space for his portico, the designer is not content to 
leave the side portions plain and square: one of them he 
feels called upon to make circular in plan, perhaps because 
he thinks a round tower on a corner looks well, and so it 
does in some cases. 


Fig. 10. 
Similarity and dissimilarity of parts. 


He might far better have let it alone, as all such efforts 
ted failures, but with both parts left square as 
—ill parts made round as at c, an excellent result 

Yet how often is this mistake made, simply because the 
■tier has never realized clearly the principle that dis- 
similarity between parts that have substantially the same 
function is always dis : »le. 

A] >art fn>m other considerations which may demand cer- 
tain departures from the rule the effect of a building is im- 
proved if all of the 
openings are of the 
same sort, all lin- 
teled, or all round, 
or all pointed. 

Compare, for 
instance, this pho- 

II c ^VVl ~\ Qli &M*^ ^ '^^Plf tograph of the Pa- 

["fijfll fj : 4at : iJ"l» btzzo Vendramini 

(Fig. n) with that 
of a recently built 
house which is an 
adaptation of it 
(Fig. 12). Notice 
that a large part of 
the beauty of the 
original is due to 
the persistence of 
the arched open- 
ings throughout ; 
of windows and doors, all with semicircular 
arch( . I', ch window has a single mullion, and the two 
spaces into which it is thus divided are also arched, the tym- 

Pig. 11. 
Palazzo Vendramini. 

Unity <>f effect obtained by prevailing simi- 
larity of semicircular arches. 



panum being filled by a pierced circle, somewhat after the 
fashion of tracery. 

There is no doubt of the beauty of the arrangement to 
all eyes, simple or sophisticated. All this has been pre- 
served in the modified 
copy up to a certain point ; 
but in the subdivision of 
the windows of the latter, 
a straight transom is used 
with a mullion below, but 
nothing above, the semi- 
circular tympanum being 
undivided. An excellent 
device and pleasing into 
the bargain in many situ- 
ations, but here it quite 
loses the charm of the 
original, resulting from 
the similarity of the small 
arches over the mullion 
and of the larger arches 
which span the whole 

It may be that there is 
some analogy between this unison of similar forms of differ- 
ent sizes and the corresponding unison between musical 
notes of different octaves. At present this correspondence 
is a mere fancy, or, at the best, a plausible surmise ; but at 
any moment the researches of psychology may show a real 
resemblance in the pleasure that the mind derives from such 
unisons, whether through the eye or the ear. 

In the same way in modern dwellings even on the most 
modest scale it is important to preserve a similar window 

Fig. 12. 
Pulitzer House. 

Loss of unity by omission of some of 
the arches. 


tment throughout. In such buildings the sashes are 
often cut up diagonally into diamonds. When this treat- 
ment is used it should be carried through all the windows, 
and should not be alternated nor varied with sashes cut up 
into rectangular lights. 

Often, under the impression that additional elaboration 
will improve the appearance, some of the sashes are com- 
plicated by the addition of curved sash bars in tortuous 
patterns, entirely sacrificing the charm which uniformity of 
treatment gives for a labored frivolity. 

Such trivial details of cottage design may be deemed 
unworthy of attention; but it is in just such trifles of every- 
day practice, as well as in monumental work, that rules of 
composition find the test of their validity. 

Another instance of the importance of similarity of parts 
is found in the common dictum that the pitch of all the roofs 
of a building should be of an equal inclination. 

This rule is not absolute, as there are places where the 
pitch may be varied with propriety, but for the present 
we may give the precept full acquiescence, leaving future 
limitations out of the question for the moment, and 
the ill results that follow divergence from the rule of 

There is, not far away, a house of the ordinary kind, 
designed by a worthy carpenter, but all the more instructive 
on that account, in which he has put the enormity shown in 
Fig. [3, a. 

Moved by a laudable desire to get a room in the attic, he 
has raised the eaves of the middle dormer, but not the ridge, 
ing a smaller dormer with a steeper pitch of roof along- 
side ot it, and the round tower at the corner has made bad 
worse, whereas with dormers of the same pitch of roof, as 
at b, we can hardly go astray; or, if we must have a dormer 



D R 


large enough, or nearly large enough, for a whole room, 
something like c may be made to look as well as possible 
under the circumstances. 

Note that as long as the pitch is more than 45 the vertical 
element in the composition prevails, and we may dispense 
with the level cor- 
nice line, but as soon 
as it is less than 45 
the level line begins 
to be demanded, 
while at just 45 ° we 
may either use it or 
dispense with it. 

The low pitch 
harmonizes with a 
continuous treat- 
ment and general 
horizontal motive, 
while the high pitch 
harmonizes with individual treatment and vertical motive. 
This is why the low pediment of the classical styles gives in- 
dividuality to the front and at the same time does not mar 
the unity of the horizontal whole. 

Whatever sort of treatment is used for the dormers, the 
same sort of treatment must be used for the main gable of the 
roof. If a rich or fantastic style is used for the main gables, 
the same sort of treatment must be used for the dormers. 
Not by any means an exact copy in miniature of the larger 
gable — on the contrary the details of the outline and orna- 
ment must be nearly on the same scale in both — but a general 
reflection in its forms of the larger. 

The doctrine that the roofs of a building must have the 
same pitch remains a dry and barren formula, to be thrown 

Fig. 13. 

Lack of unity caused by dissimilarity at a. 
Unity obtained by similarity at b. 


the slightest demands of convenience or unbridled 
, until we realize that it is but a part of a larger doctrine, 
all parts of a building that serve similar purposes, and 

Fig. 14. 
All Souls Church, New York. 

lilarity of parts is obtained by the use of the circle and semicircle 

sometimes when they serve different purposes, are 
united in a harmony of the whole by a general resemblance 
in sh 

A hemispherical dome, for instance, is felt to be ap- 
propriate in connection with semicircular arches (Fig. 14), 



Fig- 15- 

Harmonious effect of similar curves 
in plan as well as in elevation. 

and it is for this reason that the interior of most of the great 
domed buildings is more harmonious than the exterior, as the 
interior is a complete compo- 
sition of semicircular arches 
in the aisles, and in the vaults 
of the nave, culminating in 
the doubly arched dome, 
arched upward and also 
arched in plan ; while the ex- 
terior, with its cornices, pedi- 
ments, and colonnades, to a 
great extent discards the arch 
in favor of the lintel, much 
to the detriment of the unity of the external appearance. 
y^ Allusion has just been made 

/ %^ to the harmonious effect of a 

dome on account of its curva- 
ture in plan, as well as in eleva- 
tion. As a general rule the 
same sort of line that is used 
in elevation may be used in 
plan with good effect. A semi- 
circular porch often seems out 
of place unless some kind of 
circular or semicircular object 
is placed upon the flat front of 
the building to justify the curve 
of the porch, as in Fig. 15. 

Circular arches when in- 
troduced into a circular plan, 
either as a connected arcade or as separate arches, produce 
an admirably harmonious effect from the similarity between 
them (Fig. 16). 

Fig. 16. 

Unity produced by similarity of 
semicircular arches throughout. 


Equally graceful is the arcade on a semicircular plan 
from the Villa Albani, shown in Fig. 17. 

the same reason it is exceedingly difficult, if not 

npossible, to make a square dome, or any kind of tower, 

with curvilinear roof and rectangular plan, look well. To 

too, is l" be attributed the out-of-place appearance of 

curves, whether convex or concave, in a mansard roof upon 

a rectangular plan. 

( in the ether hand, dissimilarity of contiguous arches, an 
error to which engineers are especially prone, always gives 

a disjointed and painful 
appearance, as at Fig. 18, a, 
while an equally plain 
structure at b becomes 
pleasing merely because 
the three arches are all of 
like curvature. 

Even in the most mod- 
est domestic work, a line 
reflecting another in a dif- 
ferent part of the compo- 
sition, whether interior or 
exterior, will give a pleasing effect which no richness of 
adornment could produce. 

In Fig. 19, a sketch from an actual house, the large arch 
in the foreground is of the plainest, that of the fireplace is 
of ordinary rough brick, but, in spite of the almost bald 
materials, the happiest result is secured. 

s this similarity of parts that constitutes style, and 
; advantage of adherence to style by copying the 
ments of it in the past is that thereby a certain ready- 
similarity and harmony in all parts is secured without 
:ial effort on the part of the designer. 

Pig. 17. 
Arcade, Villa Albani. 

Similai micircular lines both 

in plan and in elevation. 



A more intelligent method and a really worthy intel- 
lectual exercise would be the study of any transitional style 

Fig. 1 8. 

Inharmonious effect of dissimilar arches, a, and pleasing effect of 
similarity, b. 

with the view of modifying such parts as seem incomplete or 
inharmonious, substituting others which reflect better the 
characteristic points of the style chosen. 



In the two perfect styles, the classical and the mediaeval, 

similarity of form is carried into the smallest details. 
The crocketted spire of the fourteenth century Gothic 
rch is repeated a hundred times in the crocketted pin- 
nacles below, all very closely resembling it in both pitch and 
ornamentation. And the same pinnacles reappear on 
and sedilia, in the interior, losing whatever con- 
structional use they may originally have had, and are re- 
peated merely as an ornamental form, deriving half of their 

charm from the repetition. 
In the same way the 
Gothic arch, originating in 
the construction of the 
vault, is repeated in the 
span of every window and 
as the head of every mul- 
lioned subdivision of the 
windows. Further than this, 
the form of the Gothic arch, 
cusped or plain, is used in 
panels cut out of the solid 
stone; and in the carved 
woodwork of the stalls both 
crocketted finial and cusped arch are again repeated. 

Such use of constructional forms for decoration has 
e< Kidemned, 1 »ut it is only by their repetition as decora- 
tion that they become decorative, and their use as decora- 
tion rative because nothing else will harmonize with 
tractive lines of the great vault above, which are 
: upon us as the keynote of the style. A Gothic arch 
. not be in itself so graceful as a semicircular arch; a 
innacle is not intrinsically an especially beautiful 
It is only by accepting them as the best available 

Fig. 19. 

Similarity of form of various ]>arts of 
a cottage interior. 



treatment for essential constructive features, and then re- 
peating them everywhere, playing with them, revelling in 
them, that a grand and perfect whole is made. 

The same similarity between part and whole prevails 
in the classic style, although it is not so conspicuous at first 
glance. In that 
style the keynote is 
a regular horizontal 
line of columns, in- 
cluded by horizon- 
tal lines of stylo- 
bate and cornice. 

Notice how in 
each column the 
flutings repeat the 
orderly line of the 
columns them- 
selves (Fig. 20). 
Each column is 
itself a colonnade, although a colonnade of concaves instead 
of a colonnade of con vexes. 

Notice, too, how every enriched moulding is the same 
orderly repetition of vertical lines between horizontal lines 
(Fig. 21). The egg and dart, the reel and ball, the bird's 
beak, the cymatium, are every one of them successions of 
vertical lines in orderly grouping, although the simple suc- 
cession of single columns at equal distances, which is the 
keynote both of the construction and of the decorative 
motive, is exchanged for groups of several vertical lines 
joined together by different, but never antagonistic, parts. 

The same note of similarity, and of harmony as springing 
from similarity, marks also the minor epochs, in which 
architectural art has reached a less complete perfection than 

Fig. 20. 

Similarity of succession of flutes on each column 
to that of the columns themselves. 


greal periods. Thus the Rococo ornamentation 

me of Louis XV is marked by general employment of 

louble curve, both in plan and in elevation, and, what- 

criticism may be made of such forms, the harmony of 
the result is beyond question. 

Finally, it musl be noted that the subject of the previous 
chapter, individuality and continuity, is closely connected 

with that now under 

The vertical and 
the horizontal ele- 
ments which are 
necessarily the fun- 
damental motives in 
architectural com- 
position, present in 
every architectural 
design the problem 
of reconciling things 
in their nature es- 
sentially dissimilar. 
The solution of the 
problem in classic 
lines was accomplished by insisting as much as possible on 
the horizontal and subordinating the vertical; the mediaeval 
solution is just the opposite, developing the vertical at the 
expense of tlie horizontal. 

Bu1 no system of design, no style, can entirely leave out 

ther principle, and the basis of all future developments of 

yle must be the combination of these principles in forms 

Inch permh their repetition with modification in all parts 

of the completed work. 

ii no. 11 iOL 


:: ! : : E: : : : : f ■ 



Eye) and Uart 


SDcad and TSeeZ 

Fig. 2 i. 

All minor parts ed of successions < if ver- 

lines, included between horizontal lines. 



THE next principle of composition with which we must 
deal is subordination, that is to say, the due relation 
of the parts of a building to each other with regard to their 
size or prominence. 

Hence we may define subordination to be the giving to 
each part of a building its proper 
relative importance,' the word im- 
portance being broad enough to 
cover several different kinds of rela- 
tive conspicuousness. 

An illustration will make clear 
what is meant by subordination. In 
Fig. 22 are shown two domed build- 
ings [ precisely alike, except that in 
the first, a, the dome is very con- 
spicuous, towering above the main 
building so that it becomes the most 
prominent object from every point 
of view. In the second, b, the dome 
is low, segmental, not accentuated 
by a crowning lantern. In the first, 
therefore, we say that the building 
is subordinate to the dome ; in the 
second, the dome is subordinate to 
the building. 

There arc many buildings, good 


Fig. 2 2. 

In a the dome is domi- 
nant, the rest of the build- 
ing is subordinate, while in 
b the dome is subordinate 
to the building. 


too, using the word g 1 in the usual sense in which 

rs use it, in which this matter of subordination is 

hard to determine; but even a well-designed building, it will 

und, may be improved by such a judicious accentuation 

rts as will leave no doubt in the mind of the critic as 

to which is the leading motive. 

To take another instance: in Fig 23 we have another 

pair of buildings, just alike except that in one, a, the portico 

huge affair, overtopping the rest of the building; in the 

other, b, the portico is of similar proportions, with the same 

number of columns, and in every 
way just like the large one, except 
that it is diminished in size, until its 
ridge comes below the main cornice. 
In the former, the portico is superior 
and the building subordinate to it; 
in the latter, the building is the prin- 
cipal motive and the portico is sub- 

Now there are several ways in 
which one part of a building may 
be made subordinate to others, 
of each of which we must speak 
I riefly. 

First, by a difference in -h eigh t, 
s in Fig. 24, when' the fundamental difference between the 
two buildings lies in the height to which the two front gables 

the first , a, these gal ties form ridges considerably higher 

•f the main roof; in the second, b, the ridge of the 

ain roof runs through, and those of the front gables are 

•••low it. In /> therefore the two gables are subordinate 

•• horizontal mass of the main building, while in a the 

Fig. 23. 

a. Subordination of 
building to portico, b. 
Subordination of portico 
t<> building. 



two gables are dominant and the rest of the building is 
subordinate to them. 

When planning a building some picture of the intended 
exterior is always present in the mind of the designer, and a 
clear realization of the connection 
between any necessary changes in 
plan and the changes that are in- 
volved in the elevation is essential. 

As the requirements of use de- 
mand, it may be that we are forced 
gradually to increase the width of 
the gabled projections, in working 
out a plan that we had intended 
to appear like b, in order to make 
the rooms within of sufficient size 
for their purpose. 

As the width is increased the 
height of the ridge rises. If we 
bear in mind the natural result of 
this rise as terminating in the 

Fig. 24. 

a. The main building sub- 
ordinated to the two gables. 
b. Gables subordinated to the 
mass of the building. 

rise as terminating m 
predominance of what we at first thought of as subordinate, 
we are not at a loss in dealing with the elevation. Our 
front bays have become too big to be regarded as bays 
any longer, but all the time we have had a fluctuating 
double picture in our fancy, including both the subordinate 
bays and the same bays when they become dominant, and 
we know in advance just what we shall do with them. 

Quite the most powerful implement that we have for 
giving predominance to any part is this distinction in height. 
A very few inches difference in the height of the ridges in the 
preceding diagrams is sufficient for the purpose. Height, 
moreover, is the most striking dimension by which importance 
may be added : so much so that a tower or spire that will not 


compare in bulk may, by height alone, quite subordinate 
the rest of the edifice aesthetically. In this way the Victoria 
tower, from must points of view, subordinates the West- 
minster Parliament Houses, as that of Madison Square 
Garden, in New York, subordinates the building below. 

The second way in which subordination is marked is by 
relative width. Tims many a domed building, such as the 
Taj Mahal, Fig. 25, has minarets attached to it, or near it, 

Fig- 25. 

The Taj Mahal. 

The minarets are subordinated by their slenderness. 

which, although of overtopping height, are subordinated 
by their extreme slenderness. This method, however, is 
ehiefly used where the motive is a continuous one, and 



where variations in height would be difficult to reconcile 
with the horizontal lines that necessarily characterize a 
continuous treatment. 

Thus in Fig. 26 the central pavilion is predominant 
chiefly by its greater width in comparison with that of the 

Fig. 26. 
Morgan Art Gallery, New York. 

The middle pavilion is superior to the side ones almost entirely by its 
greater width. 

side portions. It is true, there are other slight differences of 
treatment, but, as far as the massing is concerned, it is the 
difference in width that determines the subordination. 

This is, of course, a different thing from the increase in 
width that naturally accompanies an increase in height, as 
occurs in the change of treatment from b to a in Fig. 24. It 
would be possible, indeed, to increase the height of the two 
bays in b without increasing their width, and this change in 
relative importance may be effected, should the interior 
arrangements require it, by a simple change in height without 
change in width, providing other necessary changes in the 


Fig. 27. 

The gable on the left is 
subordinate in width; that 
on the right in height — an 
example of contradictory 

composition are made to unite and harmonize the very tall 
and narrow bays thai we should thus obtain with the com- 
paratively massive block of the main building. This kind 
of variation in width is merely an accompaniment of varia- 
tion in height and needs no especial 

Care should be taken upon one 
point, however, that increase in 
width does accompany increase in 
height and not the opposite; that 
is to say, that we should never at- 
tempt to make a part that we wish 
to be subordinate lower and at the 
same time broader, as such a con- 
tradiction always produces an unpleasant effect. 

A modification of Fig. 24, showing a type of this error, 
which may be called contradictory subordination, is shown 
in Fig. 27. When it is desired to subordinate one of the two 
parts to the other it is necessary to do so by changing both 
thr height and the width at the same 
time as in Fig. 28. 

But it is not of width in connec- 
tion with height that we were speak- 
ing, when we diverged for a moment 
t" speak of the two in connection 
with each other ; but rather of width 
alone as shown in the example 
given in Fig. 26. This method of 
giving importance to an element of the design is used chiefly 
in connection with compositions of a strongly marked hori- 
zontal character, in which any break in the main level 
:ce line is not desired, and the only increase in size 
t can be obtained is in width. In such a design it is 

Fig. 28. 

The smaller gable is 
properly subordinated, 
both in width and in 
he i "lit. 



usually better that the vertical parts should not have too 
strongly marked individuality, which would clash with the 
continuity that should characterize the composition as a 
whole; which suits well with the fact that a variation in 
width alone is a comparatively feeble way of marking 
subordination, just as its opposite, variation in height, is, 
as we have already noted, the most powerful. 

We now come to the third way in which predominance 
of some parts and subordination of others may be obtained, 
and that is by projection , or depth . 

It might naturally be supposed 
that we are about to repeat very 
much the same sort of thing that we 
have said ; and that what is true of 
two dimensions is true of the third, 
the greater the projection, the 
greater the importance. This, how- 
ever, is not so, except to a very 
limited extent. 

Let us consider again the two 
gabled 1 projections in b, Fig. 24, modified as in Fig. 29. 

Here the two subordinate parts, equal in height and 
width, are shown with different projections from the wall of 
the main building. It will be observed that this produces 
little impression of difference in their relative importance. 
A very considerable difference in projection may thus be used 
provided the projecting parts are of the same sort, a qualifica- 
tion which we must now explain. 

1 Gabled parts are here used for typical illustrations, because the 
pointed gable is the simplest way of indicating the individual parts, but it 
must not be supposed that these statements apply to gabled forms only. 
On the contrary they apply with equal force to all forms, and in actual 
buildings any form may be used, provided the individuality of the parts 
is indicated by some other suitable treatment. 

Fig. 29. 

Small effect of projection 
in subordinating one gable 
to the other. 


Recurring t< i Fig. -'4. we n< >te that alth( »ugb the size < >f the 
gabled bays in a and b varies, their character remains the 
same. In both a and b they arc distinct individual masses, 
in one case, b, standing in front of the main mass of the 
building, upon which they are placed as incidents; in the 
other case, a, they have become themselves the dominant 
parts, and the rest of the building is subordinate, serving 
only to c< >nnect them. 

Let us call such as these gabled portions individual 
parts, and the rest of the building, whether serving as a 
connection or, so to speak, as a background, continuous 

We may now make our previous qualification precise, 
restate our former proposition, and say that a very con- 
siderable difference in the projection of the individual parts 
from the continuous parts may be given, without making 
any serious difference in the relative importance of the 
individual parts, and without in the least subordinating one 
of them to the other. 

It is when we come to subordination of parts of different 
sorts, that is to say, to the subordination of continuous 
parts in their relation to individual parts, that projection is 

Almost all modern- buildings contain such individual 
parts, and they are found with every possible variation in 
projection, from forty or more feet, forming a deep court- 
ward, to a few inches, or even to nothing at all, a vertical line 
of quoin stones marking the point where there should be a 
projection, though this last practice is not usually successful 
and should be avoided if possible. 

As projection is the least potent means of subordinating 
individual masses to each other, it is at its best as a means 
ot separating such masses from the continuous parts; even 



a very slight projection, as we have just said, being sufficient 
to indicate what is intended. 

Thus in a, Fig. 30, the projection of the two side parts is 
only a few inches, while in b, it may be as many feet, yet a 
glance shows that these are substantially the same composi- 
tion, the sole difference being that the continuous connecting 
part is in one case recessed but a 
very little, in the other a great 
deal, from the face of the individ- 
ual masses. 

Frequent reference must be 
made hereafter, in treating of 
other subjects, to questions of 
subordination ; so that little more 
needs now to be said about it. 

It should be noted, however, 
that a very satisfactory unity of 
character may be obtained by at- 
tention to the due subordination 
of parts only, even if the other 
elements of unity be slighted or 
neglected. 'If, for instance, we 
can make one part very much 
larger than the rest, the others 

may be attached to the large one, of almost any size and 
in almost any number, providing they are relatively small. 

Thus, in Fig. 31, the minor parts are stuck on in the most 
heterogeneous fashion, without in the least disturbing the 
placidity of the great gabled mass, simply because it is 
so much bigger that the eye rests upon it alone, and forgets 
to take account of the distracting trifles that surround it. 

Observe, too, how this conclusion holds together with 
what has been said of the power of individuality to give 



No change of motive caused 
by difference in projection. 


It has been already pointed out that the pyramidal 
outline was a powerful factor in unity of effect; it was also 
pointed ou1 that the individuality of a vertical mass, as a 
tower or chimney, was sufficient to unify a varied collec- 
tion of objects at 
the foot, this again 
.-'*,-. being scarcely more 

& than a variation of 

v ^:\ the principle of a 

-f'^Ff TlY' >r ^i pyramidal outline 

, ^F'HlVwrC", fj>-"T~~m giving individuality. 

' : ' ^ ^t^M. We have thus 

I ' plSSIIil tfifejbff reached substantial- 
1 : — £!#-' ■=*—*-'' ly the same conclu- 

Fig. 3 r . sion by two different 

Unity obtained by the subordinating of all the roads; for what is 
minor parts. . , « 

pyramidal arrange- 
ment of parts but one in which a single part is dominant 
by its position as the highest, and all the rest are subordi- 
nated by their lower positions? 

What is a vertical object surrounded by minor objects 
pt a special case of a dominant object with the most 
favorablt- conditions for individuality in its shape? 

But it is time for us to take up the question of parts of 
buildings, to analyze and classify them, giving such simple 
nomenclature as naturally suggests itself for convenience 
and clearness in speaking of them. 



FOR purposes of criticism a building may be regarded as 
a solid bounded by fronts, using the word front some- 
what inaccurately to denote the sides and rear as well as 
the front properly so called, each of which must be regarded 
as a composition in itself. 

Features that dominate the whole and take their part 
in the composition from every point of view, as a dome or a 
tower, are properly regarded as a part of each front in con- 
nection with which they are visible, and are to be studied 
in their relation to it. 

These bear also a relation to the whole of the building 
as seen from any point of view; a relation which is best 
grasped by assuming a very distant view point, and studying 
the whole as a silhouette. There need be little apprehension 
that a composition which is properly arranged in every front, 
and in which proper relations exist between the fronts, will 
not compose well from any point of view and as a whole, 
as well as when each front is taken separately. 

While the oblique view of a building is invaluable, as 
the only view that shows the true relief of the different 
features, and a full conception of what will appear from 
such a point of view is essential for the proper study of the 
fronts, nevertheless it remains true that the plane of each 
front is that in which the composition must be ultimately 
worked out. 

In point of fact, although we often, indeed, usually, see 



buildings from an oblique point of view, we habitually think 
of tKem as made up of fronts, and forbear to criticise until 
a full view of each of these is possible. 

The most advantageous point of view for criticism is one 
so nearly in front that a full view of the whole front is 
obtained; and yet a little to one side, to give a proper 
sense of the depth of the various projecting and retreating 


We shall consider first the treatment of fronts as the 
only intelligible way in which to approach the subject; and as 
the way in which it is necessarily handled by practical 
ners on the drawing board. 

Le1 us go back n< »w to the second illustration of the sixth 
chapter, Fig. 23. 

In the lower figure, b, the main building is evidently 

dominant and the portico is subordinate; in the upper figure, 

a, the portico has outgrown the building and has itself 

become the dominant part, while the building is attached 

on each side as two wings. 

Again, in the next figure of Chapter VI, Fig. 24, in the 
nd diagram, b, the two gables are set in front of the 
main building and are lower than it is, and just as clearly 
in this ease, the two bays are subordinate to the main mass 
of tlie building, as was the portico in Fig. 23, b, while 
at a the two gables are dominant and the rest is subordinate, 
as was the portico in Fig. 23, a. 

In all buildings that are composed of separable parts, 
irrangements, however complex, may be pictured in the 
mind as capable of undergoing similar transformation by 
increase or diminution of the individual parts from the 
point at which they are manifestly subordinate to the main 
mass of the building to the point at which they become 
themselves the main masses, in relation to which the rest of 


the building has become merely a connective member or 
an attached wing. 

It is well to learn to regard all compositions as in a state 
of possible flux, for in the work of practical designing great 
flexibility of mind is thereby acquired. 

Much as the manufacture of a new nomenclature is to be 
deprecated, it becomes necessary for us to give names to the 
results of this increase or diminution, for convenience of 
reference in future ; in doing so let us use the simplest 
and. most descriptive of plain English words that we can 

Let us then call the individual parts primary masses, 
when they are the principal parts of the front and all the 
rest is subordinate to them; whether a single part, as the 
pediment in Fig. 23,0, or two or more parts, as the gabled 
bays in Fig. 24, a. 

On the other hand, when such individual parts are 
diminished in size until they are subordinate to the building, 
to the front of which they are then attached, let us call them 
secondary masses. 

These terms are applicable to the parts of a building 
which have the relations described, whatever their shape or 
function. Pavilions, towers, colonnades, arcades, tourelles, 
and plain walled cells, all are comprehended under the term 
masses ; and their relative size marks them as primary or 

One step further: In Fig. 24, a, the two gabled parts, 
that constitute two primary masses, are connected by a 
portion of what would become the main mass of the building 
if the primary masses were removed. Now it serves only to 
connect these primary masses. This function of connecting 
is, however, a fundamental one, for without such connection 
primary masses cease to be parts, and become merely sepa- 


ate buildings, accidentally juxtaposed, producing the same 
innoyance in the mind as a row of speculative builders' 
ch separated by its four-foot alley, and all pre- 
cisely alike. 

Let us call Ibis part which connects the primary masses a 
link, indicating indissoluble binding together as its essential 


are two other parts, the remnants of the ends of 
the building, that are left projecting when we increase the 

size of our individual 
masses from secondary 
to primary, Fig. 32, a 
and b. 

If we continue to in- 
crease the size of the 
primary masses these 
projecting parts become 
evanescent, and finally 
vanish altogether, as in 
Fig. 33, a, where the sin- 
gle mass of the portico 
has grown until it has 
obliterated the side por- 
tions, and in Fig. 33, b, 
where the double gabled 
masses have likewise ab- 
sorbed them. 

Let us call these side 

portions by the most 

natural and forceful name, that indicates at once their 

subordinate position and the possibility of dispensing with 

them altogether: let us call them appendages. 

As the individual objects may be primary or secondary, 

Fig. 32. 

Single and double primary masses, 
with connecting and attached" remnants 
of building. 



so links and appendages may be primary or secondary. 

Such a porch as Fig. 34 is to be reckoned as a secondary 

mass with secondary appendages. It is convenient to speak 

of primary links, together 

with the primary masses 

which they connect, and 

the primary appendages 

which may be attached, 

as parts of the first order ; 

and correspondingly of 

secondary masses, links 

and appendages, as parts 

of the second order. 

There is another kind 
of secondary masses to 
be noted, in addition to 
those already described; 
that is to say, parts that 
are placed above the 
parts of the first order, 
instead of projecting in 
plan from them. Such 
are ridge turrets, domes, 
when so small that they are dominated by the building in- 
stead of dominating it, and dormers, in relation to the wall 
from which they rise, although in relation to the roof behind 
they are projections in plan. 

In addition to these primary and secondary masses, links, 
and appendages, there is a third set of objects to be dis- 

This includes a heterogeneous collection of objects, such 
as doors, windows, chimneys, columns, brackets, arches, 
panels, cartouches, and smaller turrets, and dormers. 

Fig. 33- 

Single and double primary masses in- 
creased until appendages are lost. 


All of these we will rank together under the compre- 

ive term of details, a word which is generally used by 

architects for even smaller subdivisions than these, but 

which in its ordinary, untechnical sense, conveys our 

meaning well enough, 
and avoids the misery 
of coining any new 

It is to be ob- 
served that there is 
nothing hard and fast 
about this analysis 
and classification. 

A chimney maybe 
unimportant enough 
to be disregarded en- 
tirely, even as a de- 
tail; or it may be 
large enough to figure 
in the composition as 
one of the leading 
ndary masses; or a dormer may be of that intermediate 
size that makes it doubtful whether it should be ranked as a 
detail or a secondary mass. 

And sometimes an object maybe found which fulfils a 

double purpose, which is secondary in one relation and 

primary in another; or detail in one and secondary in another. 

It may be convenient at times hereafter to speak of such 

objects i if detail as parts of the third order, or tertiary parts. 

Fig. 34- 
ndary mass with appendages. 



IN addition to the qualities that we have already noted, 
the unity of an architectural composition depends upon 
the number of the primary masses that form the bulk of the 

It is unnecessary to dwell upon the undisputed fact that 
a building consisting of a single primary mass possesses 
unity in the highest degree in its elementary conception. 
Nothing can be more characterized by singleness and con- 
centration of idea, which is, after all, what unity means, than 
that which is conspicuously and unmistakably one. 

Such are a pyramid and a Parthenon, a Venetian Cam- 
panile and a Pisan Baptistery; and such, too, are a Strozzi 
palace, a Colosseum and a RadclifTe library. 

When circumstances permit the adoption of a single 
primary mass as the motive of our building, it is not to be 
lightly sacrificed by the ill-considered addition of parts, 
"to relieve it," or "to give variety." The simplicity 
and dignity that inhere in singleness of mass we may 
often endeavor in vain to find in other more complex 

When a building is of small size, like many country 
dwelling houses, the absolute plainness of the four walls 
and roof are sometimes too bald to be satisfying. " A mere 
box of a house," is the complaint; and a well-founded com- 
plaint it often is. Yet the remedy is not to cut it to pieces in 




Fig- 35 a - 
Pylons of Temple at Philae. 

such a way as to destroy this singleness of mass, but rather to 

adorn it with subordinate secondary or tertiary parts, as will 

be hereafter shown. 
But in the case 
of large city fronts 
of stores or ware- 
houses, where flat- 
ness is forced upon 
us by the necessity 
of covering the en- 
tire lot, and where 
the dignity of a 
single mass is not 
missed by lack of 
size, it is a mistake 

to attempt to make vertical breaks in the continuity of the 

horizontal lines. Such breaks often cannot be more than 

eight inches, perhaps only four 

inches, and these are quite 

inadequate to make up for the 

loss of the continuity of the 

motive that naturally belongs 

to all buildings of the kind. 
Leaving the single primary 

mass, we come to the question 

of two such masses. 

In all ages we find what are 

virtually two separate build- 
ings, each of a distinct identity, 

joined together by another 

part, used as the motive of a 

composition. p. , 

Compare, tor instance, in York Cathedral. 



Fig- 35^- 
Hotel de Ville, Lyons. 

Fig. 35, the pylons of an Egyptian temple, the twin towns 
of a church, the two pavilions of a French court house, and 
the double gables 
of many an ordi- 
nary American 
frame house, and 
note, what is really 
a very remarkable 
fact, that the two 
masses in each 
case, when proper- 
ly joined together, 
appeal to the eye 
and to the mind as 
a single whole, that 
is to say, the quality that we call unity attaches to the 
combination in a very high degree. 

Just why this should be so is a problem for psycholo- 
gists, who are daily conquering new realms. Certainly a 

motive which persists thus 
through thirty centuries 
must be regarded as an es- 
tablished principle of com- 
position and available in all 

There are two notable 
conditions attached to such 
a combination, which unful- 
filled, the combination ceases 
to be a combination, and the two objects stand separate, 
appealing to the eye as individuals only. 

In the first place there must be a visible connecting link 
of some sort to tie them together. 

Fig. 35^- 


Without this connection there arises a peculiar and dis- 
ense of aesthetic uncertainty and superfluity; 

uncertainty as to 
the point of con- 
centration of at- 
tention, some- 
what like the 
feeling of physi- 
cal doubt when 
two doorways, 
exactly alike, are 
placed near each 
other, and we 
stand hesitating, 
not knowing 
which to enter. 
There is also a 

Fig. 36. 

Comedy Theatre, Berlin. 

Lack of unity between two pediments caused by 
contiguity without subordination. 

1 ig 1 >f superfluity, of one too many, which we find in other 
situations where a strongly individual form is placed near 
another without evident 
connect ion, and without 
sufficient difference in size 
to indicate which is subor- 
dinate to the other. Such 

ase occurs where two 
pediments are put one 

ely over the other, Fig. 

or when two gables 

occur contiguously either 

at an angle, as in Fig. 37, a, h 

or upon the face of a com- Fig. 37. 

ion, as at Fig. T,J,b. Lack of unity caused by contiguity 

.,., . ' with neither connection nor subordi- 

lne same sense ot sepa- nation. 


6 7 

ration as occurs 
when there is no 
linking part to con- 
nect them, is felt 
when the recess be- 
tween is so deep 
that the link is not 
easily seen. This 
often happens in 
New York apart- 
ment houses where 

Fig. 39- 

Cherbourg Apartment House, 
New York. 

Upper part of connecting 
of sight. 

Fig. 38. 

Two masses lacking unity from the absence of 
a connecting link. 

the need of light and air 
produces a plan having 
this defect aesthetically. 
Here, in Fig. 38, is an ex- 
ample of two buildings 
precisely alike, without a 
connecting link, 1 and at 
Fig. 39 is one of a building 
in two parts with the con- 
necting link out of sight. 
Another, at Fig. 40, is much 
the same in arrangement, 
but the architect has added 
a great arch, carrying a 
piece of cornice between to 
bridge the interval; of 
which we may say that, if 

iThe lack of unity would be 
more noticeable if the obelisk 
were omitted from the vuew. 
This serves as a partial connection 
between the two domed buildings. 

link is out 


not entirely successful, it is certainly evidence that the need 
of some connection was felt. 

The second condition is that 
the two connected parts shall be 
similar. This brings us back to 
Fig. to a, which has already been 
twice condemned on previous oc- 
casions. We now condemn it again 
on the ground that while it pur- 
ports to connect two equal indi- 
viduals of equal race and rank it 
fails to do so ; and instead joins 
together an unequally matched 
pair, -of hostile race and alien feel- 
ing. It is a case of architectural 

So again in Fig. 4 1 , where we 

Fig. 40. 

Dorilton Apartment 
House, New York. 

The great arch spans an 
open court. It has nothing 
behind it, and is put in pure- 
ly to serve an aesthetic pur- 
is a connecting link. 

1 iav< • a common case of car- 
penter's architecture. It 
is easy to pass the whole 
thing as unworthy of criti- 
cism, but more instructive 
to question ourselves, in order to learn why we do not like it. 
It will be found that in addition to other, and compara- 

Fig. 41. 

The fundamental defect is the lack 
of similarity between the two masses, 
the mansard roof and the gable. 



tively venial, sins, the great crime that it commits is this 
equal union of totally different individual parts. 

Frequently, again, the same mistake is seen in the design 
of towers that^are attached to and form a part of a building. 

Fig. 42. 
Ill effect caused by the dissimilarity of the two towers. 

In order to obtain " variety" one of these is sometimes made 
entirely different from the other, with a peculiarly unfortu- 
nate result, as in Fig. 42. 

Put it down as axiomatic that, wherever contrast may 
be permissible, it is not so in any case like this ; that is to say, 
that wherever a design is composed of two primary masses 
these must be substantially alike. 


But, when we pass from the quale to the quantum; from 
the question of what kind shall these two masses be to the 

question of how large shall 
they be, we find an en- 
tirely different reply. 

It is true that com- 
positions of equal twin 
masses preponderate, as 
far as the number of ex- 
amples that can be cited 
is concerned, but there are 
enough of the unequal type 
extant, and they are of 
such pleasing appearance 
as to justify us in regard- 
ing a composition of two 
similar but unequal masses 
as admissible. 

Throughout the period 
of mediaeval architecture the westerly spires of the church 
were often made unequal. The same sort of thing is shown 
in a much more modest building, a simple entrance lodge 
at Fig. 43, and of later work there are the well known un- 
equal domes of Santa Maria della Salute, at Venice. 

In recent work the motive is used in many domestic 
examples such as that shown in Fig. 44, and in at least one 
it liable and admirable example, the office building, at 
St. Paul, Minnesota, at Fig. 45. The south front of the 
Westminster Parliament Buildings shows the same motive, 
Fig. 46. Observe that in all of these, in spite of the difference 
in size, the general similarity in form is everywhere main- 

The harmony between two unequal masses, although not 

Fig- 43- 

Entrance Gateway to Arundel 


Two primary masses of like kind but 
of unequal size. 



that which prevails between twin masses, has a parallel 
in other architectural features. It is closely connected with 
the repetition on a smaller scale of the cornice by the taenia 
of the ovolo of the capital by the neck moulding, and of the 
lower torus, in the Attic base, by the upper one (Fig. 47). 

Indeed, with the necessary modifications to adapt them 
to the larger scale, these might be turned sidewise and used 

Fig. 44. 

House at Detroit, Mich. 

Two unequal gables. 

as the motive for a facade, as the Attic base in Fig. 48. Of 
course it is not meant especially to commend the semi- 
circular outline for a gable, but only to show the analogy 
between this and such a scheme as the St. Paul office build- 
ing facade. 

Passing on from compositions of two masses, whether 
symmetrical or asymmetrical, we come to those of three 

Fig. 45- 
Office Building, St. Paul, Minn. 

■■■'! 'le of two unequal primary masses, connected by a link. 




masses. Such are found in great profusion and of all 
varieties, but we search in vain for anything corresponding 
to the two masses that is the rule in compositions of two ; 
that is to say, we search in vain for designs composed of three 
masses, all of the same size and all substantially alike. An 
example of the nearest that usually occurs is this shown in 
Fig. 49. Yet even here the central gable is noticeably wider 

Fig. 46. 
Westminster Parliament Buildings. 

The south front, that to the left, and not shown very well in the illus- 
tration, is a composition of two unequal towers, connected by the building 

and higher than those on each side. Here, at Fig. 50, is one of 
the very few examples in which the three masses are ap- 
parently precisely alike, which shows quite clearly that 
three equal primary masses .cannot be made to hold to- 
gether as a single impression. Yet even here closer observa- 
tion shows that the middle tower is somewhat smaller, 
instead of larger, than the other two ; so instinctive is the 
avoidance of exact equality. Again, a third illustration is 





Fig- 47- 

Comi K ><iti< ins of details parallel to that 
■ if two primary unequal masses. 

i, in which the three pavilions are very nearly alike, 

with many small differences which are evident on close 

examination; but such ex- 
amples are rare, and in 
perfection, that is with 
absolutely nothing to dis- 
tinguish the central from 
the side masses, none, or 
almost none, exist. 

Of the other kind, in 
which the central mass dif- 
fers from the side masses, 
the examples, as we have 
said, are innumerable. In 
Fig. 52 are seen four of 
these, arranged progressively, from a, in which the three 
masses are most nearly alike and equal, to d, in which they 
differ widely, both in rela- 
tive size and in shape. 

In c7, the three pavilions 
have but little relief, and 
there is no difference in 
height at all, the main cor- 
nice line running through. 
The general design of the 
pavilions is similar, the 
central one being marked 
by a pair of subordinate 
masses in the shape of 
rather inconspicuous cor- 
ner projections. The topping out of both sides and centre 
is nearly the same in general effect. 

In /', the disposition is substantially the same, but the 

Fig. 48. 

Analogy of outline of Attic base to a 
composition of two unequal primary 



projection of the central mass is greater in comparison with 
that of the side masses. In c, the central mass is crowned 
with a bulging mansard, and the side masses with mansards 
of rectilinear outline, all three being about equal in size. 

In d, the central mass has far outgrown the others, and 
at the same time has totally changed in character, having 

Fig. 49- 
Morris Heights School, New York City. 

Three nearly equal masses. 

become a great dome, the two flanking masses being spires 
of classical detail. 1 

Many other examples might be interpolated between 
these, constituting a regular series ; those shown are enough 
to illustrate the chief steps in the progression. 

1 Properly the illustration should be a view of the west front, in order 
that the dome might appear as the central mass, between the two towers. 


It appears that wherever three masses are used there is a 
strong tendency to break up the group of three into a single 

central predominating 

mass and a subordin; te 
gr< iup i »f two. So much is 
this the case that most of 
such designs would make 
very passable compositions 
if either the central or the 
side masses were wiped out 
altogether. This tendency 
is analogous to a similar 
tendency which psycholo- 
gists have observed in a 
rhythmic succession of 
three beats to break up 

Fig- 5°. 

('ha i 1: \r DE JOSSELYN. 

of unity of three equal masses. 

into a one-two rhythm, double measure, the first beat being 
made equal to the second and third together; and to 
the well-known tendency 
of the even-measure glide 
waltz to bec< imea two-step. 

They are like two sepa- 
rate compositions, one of 
two masses with their link 
and another of a single 
mass with appendages, 
welded togetherinto a sin- 
gle design. 

The less,' in to be learned 

from ob ion of these 

and of others of the same lg ' :,r " T 

, . , IM Tin. Bourse, Lyons. 

is this: 1 lie more , . .... 

Three equal masses, the middle one 
nearly the three masses marked by many minor differences. 



used are equal in size, the more closely they must resemble 
each other in appearance. 

In such a design as c, Fig. 52, it would be a great im- 
provement if the side pavilions could be reduced in width to 
two windows instead of three, the cornice lowered nearly to 



l, lU jSTflflf 

■flHMr'iiv aim 

1 fffe 

. 1 -0 -•* 

a. Buckingham Palace. 

c. Palais de Justice, Antwerp. 

b. Stoedel Art Gallery, Frankfort-on- 



d. St. Paul's Cathedral, London. 

Fig. 52. 

Three masses, showing increasing degree of difference between middle 
and side masses both in size and form, from a, which shows the least, 
to d, which shows the most. 

the same height as the cornice of the link, and the mansard 
reduced to a height matching the reduced width, thus dis- 
tinctly subordinating the double motive to the single. As 
it stands, with the striking difference in treatment an- 
nouncing the central mass as something different from the 


others, yet with no corresponding preponderance in dimen- 
sions, the mind is left in a state of doubt almost as painful 
as that which is caused by two equal masses of totally differ- 
ent treatment. 

On the other hand, no matter how great may be the 
diiference in size, it is always possible to use a substantially 
similar treatment. Such a design is shown in Fig. 25, in 
which the central and side masses and even the minarets 
surrounding arc topped by domes of nearly similar shape. 

It seems reasonable to speak of these latter only as 
groups of three masses, or triple groups; describing those 
wherein the central mass is very different in design, as well 
as predominant in size, as a single mass with secondary 
double masses. 

The upshot <of the whole is this : Whatever may be the 
mode of elaborating our composition; whether with sub- 
1 >rdinate masses or details, or both ; whether the motive is to 
be horizontal and continuous or vertical and individual, as a 
foundation for it we may assume any one of three arrange- 
ments of the mass, with at least two variations. The three 
motives are, naturally, the single mass and the double and 
triple groups of masses ; while the double mass may be varied 
1 >y the asymmetrical treatment, and the triple, by augment- 
ing the central mass and assigning to it such differing design 
as shall mark it as a thing by itself, dominant over the whole 
composition," and not a mere one among three equals. A 
clear conception of these five possibilities — and there are no 
more— adds to the power of the designer to an extent that is 
out of all proportion to the lucid simplicity of the facts, 
when once thus catalogued and generalized. 

As for asymmetrical triple groups, it may be possible 
to make as coherent a composition with these as with 
an asymmetrical double motive, although .hitherto this 


arrangement has been not frequently used, or perhaps not 
at all. 

If we go a step further, and inquire concerning the 
possibilities of a group of four masses, we find that there 
exists almost none, though there is a stray example here and 
there, of which not any is satisfactory ; of such the Tower of 
London is perhaps the best known. 

It seems that beyond three, the mind fails to grasp a 
group of objects as a unit, and receives in place of unity an 
impression of vague plurality. This is caused, not by in- 
tellectual incapacity to apprehend larger numbers, as is the 
case with some savages, but by a psychological tendency to 
unite one object with another on one side of it, or with one 
on each side, but not more. 

Let the designer sit down with drawing board and pencil, 
and try to make something out of four equal objects. He 
will find himself doing one of two things : either distributing 
the four in two groups of two each, as at a, Fig. 53, or putting 
a group of two in the middle with one on each side as at b, 
of the same figure ; in the first case making what is virtually 
a compound double group, and in the second a parallel 
compound double group for the central mass of a triple 

Here, in Fig. 54, is an actual example of this latter ar- 
rangement. The four towers are equal, and nearly equally 
disposed; yet the two innermost are connected to form a 
compound double central mass. Yet even here, in spite of 
the impressiveness of the four round towers, the composition 
"scatters" and does not hold together as a unit to the eye. 
A slight diminution in the size of the flanking towers will 
be found to improve the appearance of the whole. 

A parallel exists in the relation of sequence, that we call 
rhythm; in which psychology has found that beyond a 


rhythmic recurrence i >f two beats or three beats in a measure 
a separation occurs, so that a measure of four beats is divided 
into two measures of two beats each, spontaneously and 

Nor is it any more possible to construct a unit out of rive 
primary masses. As soon as we pass beyond three, we might 

Fig. S3- 
Two methods of giving unity to four masses. 

as well put a dozen or a score, for all alike lack concentration 
and oneness; tending ever, as we increase the number of 
units, to that sense of infinity that attaches to a wall forti- 
fied with towers at equal intervals of which the eye can 
discern neither beginning nor ending. 

There is a striking parallel to the grouping of masses that 



we have just discussed in the grouping used in paintings by 
the masters, in the day when composition in the "grand 
style" recognized certain types of arrangement. 

Take as an instance of a group of three the well-known 
Sistine Madonna (Fig. 55), with the Virgin central, and the 
pope and saint on each side, 
corresponding as closely as 
possible with the front view 
of many a domed building, 
with flanking towers, such 
as St. Paul's or the Taj 

This painting is also an 
example of the pyramidal 
grouping which produces 
unity of effect as much in 
painting as in architecture. 
It may be noted that the 
antithetical arrangement, 

the continuous horizontal group, is not often used in easel pic- 
tures, but is found in sculptured and painted friezes, in which 
. even distribution, and not concentration, is the effect sought. 

The parallel to the double architectural group is found in 
the other classical grouping in painting, high at the sides 
and low in the middle, as seen in Fig. 56, a " Nativity," in 
which the two figures are connected by the recumbent infant, 
which forms the link. 

Of these two groupings in combination, together with the 
single figure, most of the great paintings of the past are 

Thus Raphael's "Transfiguration," Fig. 57, contains two 
triple groups in the background, the upper pyramidally, the 
lower horizontally arranged, and what corresponds to a 

Fig- 54- 
Chateau de Villebon. 

Four equal masses, of which the 
two innermost are more closely con- 


double group, low in the middle and high at the sides, in 
the foreground, although here each member is composed of 

ral figures, as is also the link between them, i This link, 

Fig- 55- 
The Sistine Madonna. 

Irijile masses as used in painting, the central predominating, a close 
analogue to the corresponding composition in architecture. 



composed of four 
figures in the back- 
ground, in the 
original is subordi- 
nated in color, 
which cannot be 
shown in a black 
and white copy. 

It would seem 
to be a daring prop- 
osition to say that 
it is possible to 
classify all the 
buildings in exist- 
ence under one or 
the other of the 
foregoing heads 
with respect to its 
primary massing ; 
but, if allowance 
be made for many 
that are thrown to- 
gether by accident, 
a part perhaps built 
at one period, and a 
part at another, or 
for those that are 
built to give the 
impression of such 
piecemeal con- 
struction ; and fur- 
ther allowance for the innumerable factories, city dwellings, 
tenements, and the like, which are of no account as architec- 

Fig. 56. 
Nativity, Marco Palmizzano. 

The principal figures form two primary 
masses, between which the infant figure serves 
as a connecting link. 


n will be found that all buildings intended to please the 
do fall in inn' or the other of the above described classes. 

In Fig. 58 is shown a diagram of such classification. It 

All W T\^ 


i 1^ 

r? * 

K^ ■ 

Fig. 57- 

Transfiguration, Raphael. 

A composition of double primary masses below, and triple above. 



would be possible to elaborate this still further by making 
similar diagrams for buildings of various types, vertical 
or horizontal, individual or continuous, in their tendency; 



Diagram ofPkimaky Mass jug 










With, one OLpjoeTidaye 

dJ] d 


□ D 








With two appendecyes- 

Fig. 58. 

All buildings of any architectural pretension may be classed under one or 
the other of these diagrams. 

and by subdividing the third variety, those with two 
appendages, into those with symmetrical, and those with 
asymmetrical appendages; but to carry the classification 
to such an extreme would darken counsel rather than 



IN the use of secondary masses the body of the building 
is always given, upon which, as upon a background, 

the composition is to be constructed with secondary masses 

aptly disposed. 

If they are really aptly disposed, we may convert a plain 

box of a building into a perfectly satisfactory composition; 

but it will not do to stick on "things" here and there 

promiscuously: it must 
be done secundum artem. 
It will not do at all to 
put a piazza on the 
north and west sides so 
as to have shade all the 
time, and a round tower 
on the corner, because 
Mr. Jones has one, and a 
diagonal bay window on 
another, to command 
the distant view, which 
is best on the southeast, 
and mansards and tower 

Fig. 59- 

■ I try musses placed upon a building 
regardless of rules. 

roofs of opposite curvature, for variety, and a miscellaneous 
collection of windows of assorted shapes and sizes, for no 
1 "articular reason but general deviltry, Fig. 59. On the 
contrary, the secondary parts must be applied with careful 
attention to number, size, shape, and dimensions. 




The house shown in Fig. 60, for instance, began with a 
plain box, to which the conditions made it necessary to 
adhere. The oriel windows each side of the second story 
were at first intended to start from the ground as bay win- 
dows through both stories. Considerations of proportion con- 
fined them to the second story, in order that they might 
show as horizon- 
tal rectangles, ap- 
proximating that 
of the whole front. 

They are not 
required for use, 
as they command 
no view, nor do 
they enlarge the 
rooms, as they 
do not come low 
enough to permit 
the second story 
floor to extend in- 
to them ; they are 
put on solely for 
ornament. The 
veranda was made 

octagonal because the oriel windows were octagonal, and its 
roof was octagonally hipped as were those of the oriels. It 
was not carried around three sides nor two sides of the house, 
nor even along the whole front, because it was needed to 
make the third of a group of three with the two oriels. 

The main roof was studied for a long time, as the owner 
was anxious to have more room in the garret, but in the end 
gables were discarded and the roof was hipped to make it 
harmonize by similarity with those of the oriels and veranda. 

i !.._. 

mf m 

8 J 

p 1 



s '-^2 

' : A ~>— 


* -, - 



r - J 




Fig. 60. 

A plain rectangular house, adorned with three 
secondary masses, the oriel windows and the 


A further effort was made to enlarge the Little dormers, and 
to make them octagonal also, but it could not be done 

They had to be subordinated to the twin oriels, the main 
motive, and any enlargement caused them to compete in 
size with i lie latter. 

The result was most satisfactory to all, as the owner 
was quite content to put up with the loss of space in the 
-arret, in consideration of the general approval of the 
external appearance. It is a mistake to suppose that 
beauty tan always be attained without cost or sacrifice. 
h may be noted that this was the first design made with a 
conscious application of the laws of composition as now 

In this way secondary masses are to be used, by applying 
them to the primary mass or masses, which would be com- 
plete, however, without them. Thus used they may be 
applied to any parts of the first order, that is, to the primary 
masses, or to the links that connect, or the appendages that 
extend i hem. 

Corresponding with the individuality and unity that 
inheres in a single primary mass, is the sense of unity in the 
whole composition that is given by the application of a 
single secondary mass. 

Whether it be portico, or porch, or turret, or bay, or oriel, 
i >r dormer ; whether planted on in the centre, or asymmetri- 
cally on one side of the centre, or on an angle, the power of 
such a single object to give unity is attested by its general 
] >revalence. The most usual and well-worn type is naturally 
the central portico, as in Fig. 36. Of course a perfectly sym- 
metrical arrangement, and some kind of central projection 
of porch orportico, is frequent and excellent. Of deliberate 
nmetry a conspicuous instance is the tower of the 
lzzo Vecchio at Florence (Fig. 106). The appearance of 





- r r p r ' |& 




BL— ~-~ — 

; WW'-, 


this has been injured by the upper belfry, added at a later 
period with the apparent intention of making the tower the 
primary mass, while its original designer evidently meant to 
subordinate it. Single subordinate objects asymmetrically 
placed were a favorite motive of Richardson, as seen in the 
library, Fig. 61, in which the gable is a secondary mass 
compared with the building, and the low turret is still 
more subordinated and is 
attached to the gable. This 
is an interesting instance 
of the use of single second- 
ary masses asymmetrically 
placed ; and also one of the 
infrequent cases in which 
a third order of masses oc- 
curs, a secondary mass 
bearing one still further 

In the same way may 
two secondary masses be 
applied to a primary part. 

We have already shown how the two oriels are applied in 
Fig. 60. Here is an example (Fig. 62) of two turrets, 
similarly applied ; in this case to the angles, and not to the 
flat front, with a single dormer upon the latter. Another 
turret is applied to the appendages as a single asymmet- 
rically placed secondary mass. 

At Fig. 63 is a striking instance of the use of two asym- 
metrical secondary masses, although they are applied sym- 
metrically. Observe that, in spite of the difference in 
size, the similarity of character is maintained. 

Here again a single secondary oriel is placed asymmet- 
rically upon the front of the primary mass. 

Fig. 61. 
Library at Quincy, Mass. 

The gable is a secondary mass, 
asymmetrically placed. The tower 
is a tertiary mass placed asymmetri- 
cally upon the gable. 


\Vr\ often in the case of a central link between double 
primary masses, one or more secondary masses are placed 
upon the link as is the porch upon the first story, and the 
twin dormers upon the roof in Fig. 35^; but it is impor- 
tant in such cases 
that the secondary 
masses be kept 
very much smaller 
than the primary, 
in order that their 
subordinate char- 
acter may be evi- 

It is quite usual, 
too, to place the 
entrance, with its 
porch or portico, 
upon the central 
link between a 
double mass mo- 
tive ; but this is not 
essential, as the composition will hold together quite as well 
if the secondary mass, including the entrance, is upon one 
of tlic primary masses. 

Returning to Fig. 62, we have in the third turret a case 
"I t rcquent occurrence in which a secondary mass is placed 
at the termination of the appendage. The same thing is 
often dnn e when there are two appendages, similar secondary 
masses being placed upon each, and the effect is peculiarly 
graceful when they are placed upon the primary mass also, 
as is here done. 

So far the use of secondary masses is entirely analogous 
to that of primary masses, either single or double, and of the 

Fig. 62. 
Chateau d'Azay-le-Rideau. 

Two secondary masses upon the primary 
mass, and a single secondary mass upon the 



latter, either equal or unequal being available for use upon 
any part of the composition, and giving unity of effect to the 
whole, and individuality to the primary masses. 

When we come to three secondary masses we find their 
use much the same as that of three primary, the central one 
being in many cases the largest, as seen in the three dormers 
upon the central link in Fig. 145. Contrary to the rule 
in primary mass- 
ing, however, 
this is not al- 
ways the case, as 
three secondary 
masses of equal 
size may be free- 
ly used. 

Examples of 
three dormers, 
three bays, and 
so on, of equal 
size are frequent ; 
an example is 
given in Fig. 135, 
and in Fig. 64 a 
more notable 
case is shown. 
In this the three 
bays, crowned 
with fantastic 
gables, are of the 
largest size pos- 
sible for second- 
ary masses, just equalling in height the house itself. Note, 
nevertheless, that the ridge line of the house does run 

Fig. 6 3 . 
Dormitory, Princeton. 

The turrets constitute two asymmetrical secondary 


through, appearing as a very much fore-shortened gable at 
the right of the right-hand bay. It would make a more 
coherent design if the ridges of the gables were somewhat 
Lower with reference to the main ridge. It may seem like 
splitting hairs to insist on a very minute point like this in 

Fig. 64. 

House of II. W. Poor, Esq., Tuxedo, New York. 

The three gabled roofs are of the extreme limit of size for secondary masses. 

classifying masses as primary or secondary, and indeed it 
would be so, were it not that it is just this question of where 
to place the limit that so often is to be answered. 

It is true that the difference of effect where the ridge of 

gable is four inches below that of the broadside 

rool beyond, and vice versa, is infinitesimal, but if we bear 

rly in mind that the point where the ridges are of equal 
height marks the point where the front gables cease to be 



secondary masses and become primary, we have a key to the 
situation that enables us to modify our design to suit the 
circumstances, should we be called upon to change the rela- 
tions in the height of the gables in either direction. 

This illustration shows the limit at which secondary 
masses can be considered secondary and treated as such. 
Drop the horizontal ridge several feet, and the three gables 
will stand up against the sky as primary masses instead of 
secondary, and the need for increasing the size of the central 
one will at once be felt. 

The reason why it is not felt, while the front gables remain 
subordinate as secondary masses, appears to be that the 
dominant mass of the house forming a background holds 
them together, the straight horizontal ridge possessing unity 
of continuity enough in itself to connect all of the smaller 
objects that may be placed in front of it. 

When the ridges must be at the same level, as is the case 
here, it is possible to treat the three masses as primary, by 
stopping the main ridge at the ridges of the outside gables ; 
or as secondary, by letting the main ridge run through, and 
those of the gables abut against it, as is here done. 

It will be found more satisfactory to let either the main 
ridge or the ridges of the gables fairly and clearly overtop 
the other, enough to make the subordination plain, not only 
in the case of three, but of one or two secondary masses; 
indeed, the chief part of the art of composition lies in this 
definite selection and unfaltering carrying out of the chosen 

We now come to a new and totally different class of 
facts, upon which we have hitherto not touched, except by 
barest allusion ; not at all certainly in the way of explanation. 

We have just seen that the assemblage of three equal 
masses, which is unsatisfactory when the masses are primary, 


unless the central mass dominates, becomes practicable and 
pleasing when they are secondary, even though the central 
differs not at all from the others; we now note that the 
grouping of four or more masses, which is never used in the 
i if primary masses, is frequently used and always pleas- 
ing in the case of secondary masses. 

Such a row of secondary masses, for instance, as that of 
the gal iles sh< >wn in Fig. 65, is pleasing in a very high degree. 

It is so, however, in quite 
a different way from the 
groups of lesser numbers 
that we have been consider- 
ing. The sense of several 
objects forming an individ- 
ual whole is lost. The eye 
no longer grasps the num- 
ber of objects at a glance; 
the sense of individuality is 
Yet, although individuality is gone, there is still unity, 
but of a different sort — the unity of continuity. It is now 
the impression not of one gable, nor of one gable flanked by 
one <ni each side, like an escutcheon flanked by its support- 
ers, which is the fundamental impression where there are 
three, but of one building with one unbroken row of gables. 
To realize what is meant put your finger over any inter- 
mediate one of the four, and you will at once see the peculiar 
shock of discontinuity in what should be a continuous series. 
It is just the same as the disfigurement that is caused by 
tin- loss of one tooth from a row. 

And the illustration shows upon examination that the 
men of that day felt it and took steps to remedy it. Notice 
the gal ile next to that on the extreme right hand. It is the 

Fig. 65. 

Four secondary masses, exhibiting 
the unity <>!' continuity. 


only one of the four that does not occur at a break in the 
plan. Besides that the other three are placed over wide, 
multiple windows, preventing the drip of the water from 
the eaves in front of them. This one is placed boldly over 
a pier, between two windows. Sufficient evidence surely 
that it was built for "looks" only; and if it is removed we 
have the sensation of one missing that is so fatal to a 
row of anything. 

Take a single dentil out of a row of dentils, or a baluster 
from a balustrade, or an egg out of an egg-and-dart mould- 
ing, or a column out of a colonnade, and the same effect of 
stopping with a jerk will occur. This sense of satisfaction in 
an even row of things we call continuity, and distinguish it 
from the individuality which is its opposite. 

To still further clear up this question, which is a funda- 
mental one in architecture, let us take a row of three dormers 
instead of four. Cover any one of these and there will 
remain two which we know will hang together persistently 
as a pair, with no sense of a gap between them ; and if from 
a pair one is removed we have, in the one that remains, the 
sense of unity and individuality that inheres in any single 

It is only when the number of subordinate masses is 
four, or more than four, that a feeling of hiatus occurs when 
one of the intermediate members is removed. 

This feeling of hiatus, or interruption, caused by a break 
in an even row of things, makes us perceive more clearly, 
what we had before taken as a matter of course, that an 
even, unbroken row does appeal to us as a united whole; 
that, although it is without the quality of individuality, 
and the unity that springs from individuality, it neverthe- 
less has just as much unity, arising from totally opposite 



Now these properties of number are closely allied to the 
forms "t" objects that we have before noted. The horizontal 
row is closely associated with the horizontal line, and is most 
appropriately, but not necessarily used in connection with 
horizontal disposition. 

U m 'k fi »rward to Fig. i i 5, and observe that, while on both 
«.f the primary masses groups of three are used, topping out 
with single dormers, also upon the link in the three lower 
stories, in connection with the subordinate individual motive 
of tin- doorway, yet upon the appendage, that is, the whole 
rear part that is attached, four dormers, and four of every- 
thing is used, as well as upon the link when we get well up 
to tlie roof, and in close association with the horizontal line 
of the connecting ridge. 

As the continuous row is intimately associated with the 
horizontal line, so is the individual group with the pyramidal 

outline that we have spoken 
of as the characteristic indi- 
vidual shape. Observe how 
naturally the group of three 
at a, Fig. 66, fits into the tri- 
angular gable, the tendency 
to make the central one of 
a group of three the largest 
having much in common 
with the pyramidal senti- 
ment ; while a single one (6) 
will do quite as well, two (c) 
not quite so well, and four 
(J) not at all. 

We may also note that 
the reason why a moulding does not look well when cut off 
square, instead of being returned, is due to the sudden stop- 

Fig. 66. 

Adaptation of individual groups of 
to primary m.-iss of individual 



page of continuity, a sense of sudden interruption, somewhat, 
analogous to the feeling that is produced by a missing one in 
a row of objects. 

Why it is that a continuous arrangement of primary 
masses cannot be used is not so apparent as the fact that it 
cannot. The reason would seem to be that a more con- 
spicuous connection than any link that can be devised is 
needed to hold them together ; or let us say that, when the 
link is made important enough for this office, it ceases to be 
a link and becomes the dominant mass, reducing what had 
before been primary masses to a secondary position when 
compared with itself. This is also true in the case of three 
primary masses of equal size, an arrangement which always 
calls for some predominance of the central mass to give the 
needed unity. In Fig. 65, and also in Fig. 64, imagine the 
roofs of the main buildings between the gables removed, 
so that the gables stood up stark against the sky, primary 
masses beyond all question, everything else being sub- 
ordinated to them. The lack of unity, the feeling of dis- 
location which then occurs, comes from the lack of sufficient 
connection between them; and, if we again add the roofs, 
little by little, beginning with a low ridge and gradually 
raising it, we shall find that just where the roof begins to 
be sufficient to hold the gables together is the point where 
the ridges are at the same level. 

In passing from primary masses to secondary masses and 
details, we find that less and less provision for a linking part 
is required. With primary masses a link is always required ; 
with secondary masses, the connection that comes from the 
mass of the building behind them is usually sufficient, al- 
though at times a link of some sort may be used (an instance 
is shown in Fig. 67), in which the framed and shingled seg- 
mental balcony serves as a link to connect the two gables, 


while in the case of details no connecting part is required, bu1 
they may be placed anywhere and will be sufficiently con- 
nected by tin' preponderant primary and secondary masses 
up<>n which they occur. 

The conclusion is that for both primary and secondary 
masses, either one, two, or three at a time may be used, with 
the unity thai attaches to individuality, except that in the 
ease of primary masses there must be some predominance 

fel i £±. \ JJLI ' 

Fig. 67. 

Two unequal secondary gables, connected by a flat arch, as well as by the 
mass of the house upon which they are placed. 

in the size of the central mass when three are used. In the 
case of subordinate masses a very slight, quite unnoticeable, 
excess of size in the central mass gives a certain indefinable 
grace, and counteracts its tendency to seem smaller than the 
other two. 

But while more than three primary masses may not be 
used, four or more secondary masses or details begin an 
entirely new kind of harmony of number, and may be used 
in the proper place with perfect freedom. 



THE word details is here used in its colloquial sense to 
designate such minor architectural objects as win- 
dow and door openings, panels, niches, columns, arches, and 
the like, which may be applied to the masses of a building, 
whether primary or secondary, rather than in the more 
technical sense, in which it denotes still smaller parts, such 
as dentils, capitals, and brackets. 

Using the word in its first sense it will be found that the 
number of details that should be used is governed by pre- 
cisely the same laws as those of secondary masses. It will 
hardly be worth while to repeat the same things that we 
have said of secondary masses. A few instances will suffice. 

In the first place, as in the case of secondary masses, a 
single detail properly placed will often be sufficient to give 
unity to a whole composition. 

In Fig. 68 is shown a building of considerable size, of 
the horizontal continuous type in its composition, having 
from necessity, to all appearance, what is a rather unusual 
condition, two entrances of equal importance. Now two 
such objects without closer connection, and with their 
individuality strongly marked by pediments or otherwise, 
are apt to have an appearance of unnatural separation, a 
fault which is commonly known as "double composition." 
To avoid this the designer has placed a panel between them 
carved in low relief with a figure subject : the effect of this in 
"pulling together" the whole building is remarkable. 


Fig. 68. 
St. Stephen's Parish School, New York. 

The carved panel forms a point of concentration, and gives unity to the 




Another instance is found in the doorway and the 
escutcheon above it of the Farnese palace, Fig. 108, although 
there are no antagonistic twin doorways to be harmonized ; 
and examples abound everywhere of a central doorway as 
the single detail that unites a whole front. 

In most compositions of two primary masses, whether 
they are equal or unequal, a single object, either a secondary 
mass or a detail, is placed upon the link between them, 
usually, although not necessarily, at the centre. 

Thus, in Fig. 69, the panel over the central archway, as 
well as the archway itself, gives a sense of unity to the whole 

As in the case of secondary masses, details either singly, 
or in groups of two or three, that is to say, in all combinations 
that possess indi- 
viduality instead of 
continuity, are nat- 
urally and appro- 
priately used upon 
the parts of a build- 
ing which them- 
selves possess indi- 
viduality, that is to 
say, upon the pri- 
mary masses and 
secondary masses. 

Turn back to 
Fig. 5 2 , and in three 
of the illustrations, a, b, and c, the pavilions have three win- 
dows in each story, and the same is true of the example at 
Fig. 5 1 ; while at Fig. 3 5 , c, there are two windows on each 

It is essential, if two are used, that they should be alike 


Fig. 69. 

effect of single details, the archway 
and the panel above it. 





- & 



1 K'TOKL'.liUM.'ll 1 







in design ; and if three, either that all should be alike, or the 
two outside alike and the central one different. 

This means that if the first draft of the plan indicates 

something like a, Fig. 70, a 
pair of doors opening from 
a hall upon a balcony on 
cue side, and two windows 
in a room on the other side, 
such a combination is not 

If door there must be, 
the two windows must be 
put together, and a pair of 
doors substituted for the 
single one, matching the 
double window in general 
appearance, the stiles of the 
doors diminished as nearly 
as possible to the width of 
those of the windows and 
a group constructed of two 
like objects instead of three 
unlike, as at b ; or the three 
may be retained and the 
doorway made single, both 
door and windows other- 
wise being alike, as at c. 

The arrangement of city 
Ik ruse fronts, with the door- 
way on one side, is so familiar that we have lost sense of its 
ugliness. Once in a while a similar case occurs in a second 
story, as in this case, or some other unusual place, when its 
awkwardness becomes striking. 







^ s . 

'1 t\'', t 'i t '»»»'..'\«K. , t;',f',,"u , i«y,flft?t. , sz f ,»M«i 1 











'l i'ft.'it't .'■.."■."..'»! « ii.'i ;t iii>«".;wf. 'i a.".y..'tt'.K.';.'i.' , .:i I 





Fig. 70. 

■ (. Improper, b and c, proper treatment 

of details. 



Passing from definite and individual numbers, which 
cannot exceed three, we find the same results with four or 
more details as in the case of secondary masses, only in a 
greater degree, because, while it is seldom possible to put 
more than four secondary masses upon a building, on account 
of the limits that we must set to its total dimensions, it is 
possible to put on details by the dozen or score. 

Now, just as the individual groupings of one, two, or three 
details are naturally associated with the vertical, individual 
parts of a composition, so 

the continuous succession 
of four or more is associ- 
ated with the horizontal 
line, and the connecting 
parts of the composition. 

This is illustrated, too, 
in the minor details, us- 
ing the word technically, 
which are subject to the 
same rules of composition 
as the building itself. 

In an Ionic capital, as 
an illustration, the egg- 
and-tongue moulding and the horizontal lines of the bead 
and abacus perform precisely the same function in connect- 
ing the two volutes and eyes as is performed by the arcade 
in connecting the two pavilions in Fig. 35, c. 

Indeed, if we turn the capital upside down we convert it 
into a sort of crude resemblance of a double primary mass 
building, by those changes only that are needed to make it a 
stable construction and to adapt it to its uses, as at Fig. 71, 
which, rough and impossible as it is, clearly shows the 
analogy between the continuity of the connecting objects, the 

Fig. 71. 

Analogy of composition between an 
Ionic capital, and two primary masses, 
connected by an arcade as a link. 


, and darts in one cast.', the arches and columns in the 
other, and their association with horizontal lines. 

This association with horizontal lines seems to be 
essential to the full effect of continuous numbers of objects, 
whether secondary masses or details. Not only do we think 
of a row of objects as a horizontal line in itself, but, with- 
out some actual horizontal line, the row of objects remains 
dislocated, as a plurality of separate things, not as one row 
of many units. 

Half a dozen columns remain half a dozen columns, each 
asserting its separate individuality almost as much as if there 
were only two or three, as long as no entablature is placed 
above them, Fig. 72, b. It is only when the horizontal 
lines of cornice and stylobate are added that we entirely lose 
sight of the separate entity of each and conceive the whole 
as a single colonnade, as at a. 

The characteristic of such a succession is its even and 
unbroken continuity. 

A noticeable difference in the spacing of the objects, 
ci ilumns, windows, or whatever they may be, or a noticeable 
variation in the design of one of them, still more the omis- 
sion of one, breaks in upon the smoothness of the series 
and produces an unpleasing result. It is for this reason 
that coupled columns are to be avoided except in an even 
series of couples. If used in single pairs at the ends of the 
colonnade, or if interpolated in a series of single columns, 

result is apt to be unsatisfactory. For the same reason 
the use of three columns at an angle in a colonnade of 
coupled columns is to be avoided. 

Whatever the objects may be, uniformity of succession, 
where that is the quality sought, must be unbroken. 

In Egyptian work the long lines of painted and incised 
figures, all of exactly the same shape, all marching in ex- 




<Mi *s. 

actly the same attitude, were not merely pictorial, they per- 
formed a part in the decoration analogous to that of the 
columns, metopes, and carved mouldings in Greek architec- 
ture. Indeed, it is very much 
the same sense that is appealed 
to in the ordered alignment 
and marching of a column of 
soldiers to-day. 

The converse also is true. 
As horizontal lines added to 
a continuous succession of ob- 
jects gives unity to them, so a 
horizontal line, which by itself 
lacks unity, receives this quali- 
ty when a succession of objects 
is added to it. 

The primary masses of a 
building, as we have seen, may 
in their skyline be either pyra- 
midal or horizontal. 

If they are the latter it be- 
comes almost essential to con- 
firm their continuity by plac- 
ing a succession of brackets, 
mutules, dentils, or other ob- 
jects upon them. This is the 
reason that in the classical 
styles, whose keynote is the 

horizontal succession, such decorations are used ; while in 
the Gothic styles, using the term broadly, where gables and 
spires mark the individual parts, anything but a sparse ball 
flower is unusual ; nothing certainly to compare with the 
consoled and mutuled cornice of the classic. 

Fig. 72. 

Enhanced continuity given to a 
row of columns by the addition of 
continuous horizontal lines. 


Fig. 73. 

Disastrous effect of interrupted 
a mtinuity. 

For the same reason all interruptions of a horizontal 
line and of a succession of objects arc dangerous, unless we 
are prepared to deal with the parts into which we split it 
up as separate individuals. Tims in the treatment of cor- 
nice shown in Fig. 73, we almost lose sight of the horizontal 

feeling and naturally separate 
it into a group of three shelves, 
each with a pair of consoles. 

Continuous rows of details 
may be made to serve very well 
in place of a strong horizontal 
line, where other considerations 
forbid us to make our cornice 
as heavy as we might like. 
Such is the function of crenel- 
lations in buildings where there is no military occasion for 
them, and that of the crowning balustrade in many a mod- 
ern building. Often again a whole upper story is treated 
as an attic, but is cut up with pilasters and flnials, the 
cornice below not being sufficient to unify the mass of the 
building beneath it. 

'Flic whole question of continuous ornament has wide 
limits. Extending it for a moment to the field of flat de- 
or design in general, w r e find it a very marked charac- 
ter of all kinds of borders, in connection with straight lines, 
not limited in this case to horizontal, /it is probable, too, 
that it is closely connected psychologically with the surface 
designs called powdering and diapering, in which the con- 
tinuity is superficial rather than linear. Take the pattern 
commonly called the polka dot, an even arrangement of 
circles of one color on a ground of another. The circles in 
themselves are not sufficiently beautiful to account for the 
pleasure generally felt in this pattern, nor is the beauty of 


the color the cause, as, in this, nothing is more generally 
satisfactory than the plainest black and white. It appears 
to be nothing but the even succession in all directions of 
things exactly alike that is the source of gratification. 

In architecture, however, the use of continuous orna- 
ment, or arrangement, in a vertical direction is limited by 
the possibilities of materials, and by the continual intrusion 
of the intellectual questions of construction and utility upon 
the pure sentiment of beauty with which the aesthetic sense 
theoretically must deal. We cannot spot a white wall with 
circular polka-dot windows arranged quincunx fashion. 

Two principal rules may be laid down for the architec- 
tural use of details in number four or more, that is to say, 

In the first place, when the treatment is horizontal, and 
the horizontal dimensions of the front exceed the vertical, 
and there is no need for breaking up the front into pavilions, 
and a plain straight skyline seems the natural thing, in such 
case the even succession of windows, as in the Farnese, and 
many other Italian palaces, is the proper and inevitable treat- 
ment. When a continuous treatment has been adopted, 
refrain from breaking up the regularity of the succession on 
pretense of fortifying the abutments as in Fig. 74, or any 
other plea of constructional necessity, as such devices no 
effort of the unsupported intellect can excuse. 

In the second place, where a flat front and straight sky- 
line are not available ; where the design must be divided 
into primary masses, one, two, or three, with their essential 
links and possible appendages, the proper place for the con- 
tinuous treatment is upon the links or appendages, and not 
upon the masses, although where the masses are of a strongly 
marked horizontal character and the level cornice is car- 
ried through, it is possible to give them also a continuous 


tment, in the disposition of the details as in this illus- 
tration (Fig. 75), of the garden front of the Pitti palace, in 
which each of the two masses has five openings in each 


The great difficulty about architectural design is the 

that there can be no single solution for a given arrange- 

Fig. 74- 
The change from segmental to semicircular arches destroys the continuity. 

ment. The ways of doing things are many, all good if 

] >r< >] >erly d< >ne ; but as, when a certain change is made, others 

•nc necessary, the problem is how to make such simul- 

ous modifications in all other parts as shall secure a 
harmonious whole as the result. 

Fig. 1 r ^ is an admirable specimen of the appropriate use 
"f individual groups of details, and also of 'continuous ar- 



Fig- 75- 
Pitti Palace. 

The masses, as well as the link, are treated 

rangemetits of them. On each of the two pavilions with 
their high peaked roofs and marked individuality the win- 
dows are placed in groups of three, terminating above in 
a single window; the link also has three, as a central door- 
way was required, 
not only for utility, 
but as a single uni- 
fying detail ; never- 
theless, immediate- 
ly below the hori- 
zontal roof line of 
the link is a row of 
four dormers, very 
much adding to the 
unifying effect of 
that line. 

Upon the ap- 
pendage the windows are placed in rows of four, denying 
individuality to this and concentrating it entirely in the two 
primary masses. Notice, in passing, how carefully the roof 
line of the appendage is subordinated. Not only is it made 
much lower, as is proper, but the very cresting upon it is 
smaller than that upon the ridge of the link. 

That the mediaeval architects realized the importance 
of the continuous horizontal line in the link, as well as the 
difficulty they encountered in reconciling it with the verti- 
cal gable, is seen in many church fronts, as at Paris (Fig. 
102), in which the horizontal arcade quite masks the gable, 
at Amiens, where the gable is only partly obscured, and at 
Tours, where they contend for supremacy. 

Finally, it is possible, by suitable arrangements of de- 
tails, both individual and continuous, to delineate, so to 
speak, upon a flat wall whatever composition, analogous 


'...., ■'':..-.:■. ;.o.;;. 


to those of mass, link, and appendage, may best suit the 
matter in hand. 

The most available details for the purpose are the win- 
dows and doors which are essential parts of most modern 

structures, and these may be 
disposed in an endless variety 
of combinations, that par- 
ticular motive being chosen 
which best suits the internal 

Figs, 76-79 are a series of 
diagrams suggesting the sort 
of thing that may be done. 
The first (Fig. 76, a) corres- 
ponds to a single primary 
mass, with two asymmetrical 
appendages. The doorway, 
which may be enriched in 
such style as will match the 
rest of the ornament, consti- 
tutes the single object ; while 
the windows on each side 
take the place of the appen- 
dages. Above, the row of bull's-eyes, either with or with- 
out connecting enrichments, forms a continuous line, which, 
ther with the cornice, connects the whole. 
At /> is the same motive slightly modified. The frieze 
of bull's-eyes is abolished and only the doorway and win- 
dows remain, suggesting as before a mass with appen- 
dages. Such a use of a single preponderant detail may be 
made just as effective, and may conduce just as much to 
the unity of the whole composition when placed asym- 
metrically, as in both of these sketches, as if it were duly 

Fig. 76. 

Disposition of details suggesting 
!.■ mass with two asymmetri- 
cal appendages. 


^W?QT=-j»Q{vt 1=3 O T"--.: O H3- O ~> =.< Cm « O r 








g;Q i&C O ty.Q ?>q ;>j r-_r Oi^o^OiSrQv 

centred, perhaps at the sacrifice of important requirements 
of utility. 

An arrangement of three large windows is shown at Fig. 
77, a, corresponding with the three secondary masses in 
Fig. 64. The smaller win- 
dows on each side of the 
main group in a case of this 
kind must be ignored as 
much as possible ; very of- 
ten the composition would 
be improved if they could 
be omitted entirely. Where 
that is not practicable, it 
is always possible, by mak- 
ing the reveals slight and 
the treatment bald, to ren- 
der them almost unnotice- 
able in the general scheme. 

A modification is at b, 
in which the triple motive 
has become a double one, 
the two smaller arched win- 
dows serving as a connect- 
ing link. This use of smaller 
objects as links between the 
larger details that constitute 
the group, is a characteristic 
of the treatment of details, 
and of great utility in prac- 
tical work. 

Another instance is at c, 
in which the three large windows of a are connected by 
smaller windows between. 

I U U i(_ — |i,^j ^ '. 



~-* Qr-3 Q??3 Q^-i O -* ?<Q K^Q?I,; C^O^on 





Fig. 77. 

Disposition of details suggesting dou- 
ble and triple secondary masses. 

i i a 



At Fig. 78, </. is a simple one-story building, the two large 
windows forming a double group. The four smaller win- 
dews between constitute the link and the two external stand 

in place of appendages. The 
whole corresponds to such a 
composition of primary masses 
as that at b, in which the 
matching of the parts, each to 
each, is indicated. 

An entirely different type of 
np^^uiH building is shown at Fig. 79. 

This is the many-storied build- 
ing of modern times. It is here 
treated by combining the win- 
dow openings in four stories 
into two large arched openings, 
which it seems anomalous to 
call details, and yet they cer- 
tainly are not masses in any 
Whatever we choose to call them, it is a composition 
of two objects, the smaller windows in the central and flank- 
ing piers being ignored entirely, and suppressed as much as 
possible. In the two lower and the two upper stories the 
openings are thrown together as continuous rows, in har- 
mony with the horizontal treatment of the building. 

Another class of details consists of those that are used 
to divide the parts of a building into minor parts by means 
of vertical lines; such as buttresses, pilasters, vertical lines 
"I quoin stones — chainages, as the French call them — and 
similar devices. 

Inasmuch as these are virtually lines of separation be- 
tween parts, it is the parts which count in the composition 


Fig. 78. 

Analogy between arrangement 
of details in a and primary parts 
in b. 



rather than the lines. By means of such details any wall 
surface may be cut up into three parts where individual 
treatment is required, and these may be all of equal size, as 
shown in Fig. 51, in which each story of each pavilion is 
thus divided by pilasters, or into more than three parts 
when continuity rather than individuality is sought. It is 
noteworthy that a wall which is divided into two parts by 
vertical details does not give the unity which usually in- 
heres in a combination of two objects, for the reason that 
a vertical line appears as a division rather than a connec- 
tion, and such an arrangement can be used only where con- 
tinuity is sought and there is 
not room for four divisions of 
the surface. This appears in 
the same illustration (Fig. 51) 
in the division of the links into 
two parts. In similar situations 
a division into two parts pos- 
sesses a stronger effect of con- 
tinuity than a single part or 
three parts, and is to be pre- 
ferred if more than three cannot 
be compassed. 

Vertical details are often 
used to accentuate the outlines 
of masses whether primary or 
secondary, as the buttresses at 
the angles of a Gothic tower, 
which are not required con- 
structively ; or the pilasters, sin- 
gle or coupled, at the angles of a pavilion, in much the 
same way that a cornice is used to outline the top, or a water 
table the base of the building. In the same way chainages 

Fig. 79. 

Disposition of details to suggest 
a composition of two secondary 


of quoins are often used at the angles of a building or of its 
principal divisions. 

Sometimes a compound subdivision is made by means 
of such details, as in Fig. i i . in which each story is divided 
unequally into three parts by coupled columns, and the 
middle <»ne of these is again divided into three by single 

For continuous treatment, too, vertical details, such as 
columns or pilasters, are frequent; in which use they are 
subject to the same rule as other continuously arranged 
objects, that the series must be even and unbroken; and 
this applies even more forcibly to these than to details of 
other character, for whereas the latter often seem to almost 
spontaneously group themselves when irregularly placed, 
any noticeable irregularity in the spaces into which walls 
are continuously divided by vertical details is sure to be 

Nor can any attempt at asymmetrical division be made 
with such details. The division into two parts must be 
into tw< » equal parts ; and that into three parts must at least 
have the two outer parts equal to each other. 

I )isregard of this rule is frequently to be observed where 
the principal but lesser front of a building is divided into 
three parts by quoins, and an equal space is cut off on the 
return by a like line of quoin-stones, leaving the rest of the 
long side undivided, with a very unsatisfactory result. 



WE have spoken so far of the various parts of buildings 
formed by vertical lines of separation. 

Masses and links and appendages are distinguished 
either by a difference in projection, which means a vertical 
line of shadow, or by a difference in height, which means a 
vertical line at the point where the difference in height oc- 
curs, or by both. They are also subdivided by means of 
pilasters, buttresses, engaged columns, and such other de- 
vices for obtaining vertical lines, as described ■••in the pre- 
ceding chapter. 

Whatever be the sort of building under consideration, 
whether a simple four-walled in closure, or a highly complex 
assemblage of parts of the first order, it is almost always 
cut up into horizontal parts in one way or another. 

There are two ways in which a building may be so di- 
vided, which are strictly analogous to the ways in which it 
may be divided into parts vertically, but of inverse impor- 

The first is by the advance and retreat of different parts, 
that is, by either setting back or by overhanging the upper 
stories, producing either a shadow or a line in construction ; 
the second by the delineation upon the surface of the front 
of lines of demarcation, at the points where they are re- 

The first way is exemplified in such towers and spires 
as are built in gradually diminishing stages, also by porti- 


coes, aisles, verandas, or arcades, that advance in front, as 
does the narthex of St. Mark's at Venice. A similar scheme 

was carried out on a large scale in the pyramidal build- 
ings i >f many stories, each setting back from that next below, 
which were built in early times by the Assyrians and Mexi- 
cans. This method is little used now because of the very 
small effect which is obtained, at great expense of construc- 
t i . m, ami serious loss of space internally. The effect, too, of 
a line obtained by retreat of the part above is almost noth- 
ing, except where the gradually tapering silhouette of the 
outline can be seen. This can only occur in a tower, for 
in an extended front the profiles are too far apart to be 
grasped by the eye in one picture, while upon the face of 
the front, the retreat of the upper part produces no strong 
sha'lnw, and is barely noticeable except by its bad effect in 
foreshortening and cutting off the superstructure in any- 
near view. Architects are familiar with the difficulty of 
handling the veranda of a country house, when it extends 
along the whole front, and are justly inclined either to 
shorten it, so that it may appear as a secondary mass, or to 
reduce it to a loggia, or to dispose it at the end of the 
front, where it may take its place as an appendage. 

Add to this, that any retreat of the upper walls means 
an almost impossible cutting up of the plan below by the 
unavoidable continuation of the upper walls downward 
through the first story to a firm foundation, and the reasons 
are seen to be ample why this way of making horizontal 
divisions is not much used. 

As for advancing the upper part, so as to cast a heavy 
shadow, no doubt this would be most effective; but here 
again, even more forcibly, constructional difficulties forbid. 
It is impossible to support an overhanging wall of masonry 
if the overhang is more than trifling. Where overhanging 


stories have been built they have been of timber construc- 
tion, supported by projecting timbers, upon which it would 
be manifestly impossible to carry a heavy wall. Neverthe- 
less the effect of the overhang where it exists is admirable, 
and if such overhanging structures were generally available 
the method would no doubt more than justify itself artis- 

It is possible that with modern engineering construction 
of steel such designs, whether with overhanging or retreat- 
ing stories, may become practicable in stone or brick, as the 
weight of these is unimportant where metal supports can be 

For the present, however, we may confine ourselves to 
the other method of obtaining horizontal division by draw- 
ing lines upon the surface. 

The ordinary method of drawing horizontal lines in 
architecture is by means of mouldings. 

Hitherto mouldings have been talked about and written 
about chiefly from a constructional point of view, and they 
who regard correct construction as the one essential of good 
design have thought it necessary to justify the existence of 
every stringcourse by some supposed utility, and have freely 
denounced the useless cornices of classically derived styles. 

No doubt where an external offset in a wall occurs, a 
stringcourse is an admirable expedient to prevent the ir- 
regular weather-staining of the surface below, and the decay 
of the wall from the absorption of water at the horizontal 
shelf which would naturally occur. Artistically, however, we 
have nothing to do with such considerations, except to refrain 
from building anything that raises a question in the mind as 
to its stability, an intellectual doubt, which, although quite 
unrelated, might overshadow any aesthetic pleasure. 

It is fortunate that on one point artistic and construe- 


tional requirements coincide, that is, upon the desirability 

of an "undercut" moulding for cither stringcourse or 

I onstructionally the "undercut" is quite essential to 
properly shed the water, and the excellent results from its 
use, and distressing appearance owing to neglect of it, are 
matters of daily observation. 

Artistically the undercut is quite as necessary, as the 
only way of intensifying the shade which it is the object of 
the moulding to produce. 

For, from a designer's standpoint, a moulding is but a 
means of drawing a Mack brush mark across the face of the 
wall. From the eighth-of-an-inch arris upon a fillet, which 
is felt rather than seen, to the eight-foot cornice projection 
of an Italian palace, all are but the charcoal streaks with 
which we draw our design. Sometimes indeed the streak 
is narrow, dark, and uninterrupted; sometimes it is lighted 
up by speckled enrichment, sparkling in full light, or glow- 
ing in the half gloom of the shade. 

Or it may be that we draw our line as a band of enrich- 
ment only, as frieze or dado, either alone, or in connection 
with mouldings or cornice, which intensify the effect. 

Whatever enrichment is used it is most effective in such 
situations if it includes the element of repetition, which as 
we have seen gives the impression of uninterrupted con- 
tinuity, and is inseparably connected with the horizontal 
motive. Thus the dentils, mutules, and modillions of the 
classic, the dogtooth and billet of the round arched and the 
ball (lower of the Gothic styles were used. 

< Observe, too, that the more perfectly developed the hor- 
izontal motive in a style, the more clearly the repetition 
« >f the continuous ornament is marked in its enriched mould- 
ings, and in addition, the more distinctlv are the lines of the 



enrichment perpendicular to the horizontal line < >f the mould- 
ing. Thus in the Greek, the most perfect development of 
the horizontal treatment, we have in the Doric the vertical 
lines of the triglyphs in regular recurrence, forming a band 
of continuous enrichment entirely analogous to the vertical 
lines used in the enrichment of mouldings. Of mouldings 
proper, the dentil, the egg and tongue, the bead and reel, 
the line enrichment of the bird's beak, the fret, almost every- 
thing in fact, in the earlier Greek styles, is characterized 
by vertical lines. As the naturalistic idea is introduced 
with the development of the Corinthian capital, the mean- 
der, the rope moulding, the laurel leaf torus, and others, 
constructed on lines that are not vertical, become more fre- 
quent, until, in the Gothic, with a strong tendency to plain 
mouldings entirely, such continuous decoration as is found 
consists of wavy lines with recurrent leafage, and with but 
few lines perpendicular to the moulding anywhere. 

Sometimes, when a moulding cannot be used, or cannot 
be made heavy enough for the desired effect, the band of 
successive objects may take its place, in part or entirely. 
Such recurrent objects are the military, or quasi-military, 
machicolations of the Gothic, the swallow-tail crenellations 
of the Venetian, the anthemion crestings of the horizontal 
styles, and all balustrades upon the face of a building. 
Sometimes even a line of windows may play its part as a 
horizontal line in the composition, especially when, in attic 
or mezzanine, they can be reduced to bull's-eyes, and tied 
together by enrichment of the surface between. 

In our modern buildings of excessive height, whole sto- 
ries, with cornices above and below, complete compositions in 
themselves, are often used as mere lines, to separate the base- 
ment of four or five stories from the shaft of ten or twelve, 
or the shaft from the superstructure of six or seven stories. 


Whatever be the motive adopted, whether a vertical 
one, with a tendency to tall and comparatively narrow 
masses, and a general accentuation of heavy vertical lines, 
or the opposite horizontal motive, with the masses broad, 
flat, and level, there will be necessarily some lines, both hor- 
izontal and vertical, in the composition. It must be borne 
in mind that one or the other set of lines must predominate. 
A heavy cornice breaking around many deep projections is 
to he avoided, as well as strong vertical lines in connection 

with a design whose leading 
motive is its marked hori- 
zontal crown. 

Of the former error the 
west front of Westminster 
Abbey is a notable instance, 
in which the classic procliv- 
ities of Wren induced him 
to build semiclassic cor- 
nices upon a Gothic front. 

The most frequent use 
of horizontal mouldings is 
naturally for the cornice 
and base of the building. 
Whatever be the style, 
some kind of moulding, whether the heavy classical shelf, 
< ir the simple terminal stringcourse, seems to be needed at 
the to]) of the vertical wall; while a base of some sort, in 
the form of steps, or the moulded dado, of greater or less 
height, is almost as universal. 

Between these the wall surface itself may be cut up by 
mouldings in various ways. 

Most frequent is the division into two substantially equal 
parts. Here is a typical example (Fig. 80). Another in- 

Fig. 80. 
Banqueting Hall, Whitehall. 

Horizontal subdivision into two equal 



r^Cffij ■ ■• 


ifiv jggf 



ft ?9 

Fig. 8 1. 
Bourse, Marseilles. 

Horizontal subdivision into two nearly 
equal parts. 

stance is the familiar library of St. Mark, at Venice, and very 

many other buildings are arranged in this way. There are 

others in which one part or 

the other predominates so 

slightly that the impression 

is that of general equality 

of height, although the im- 
portance of one or the other 

may be accentuated by the 

enrichment. An example 

is given at Fig. 8i. 

Besides these there are 

innumerable specimens in 

which the division is into 

two parts, of which one or 

the other predominates so 

much that its excess in height is manifest or conspicuous. 

Sometimes it is the basement that is subordinate, with a con- 
siderably higher main story 
above, as in Fig. 82, and this 
is frequently carried to ex- 
tremes, especially in build- 
ings of the vertical type, 
where the horizontal base- 
ment, marked by a not too 
prominent line of mould- 
ing, seems to bind the whole 
together, when any addi- 
tional lines nearer to the 
top might antagonize the 
general vertical tendency. 
The opposite way of using unequal double horizontal 

division is sometimes seen, the lower part being made much 

Fig. 82. 

Louvre, Paris. 

Horizontal division into two unequal 
parts, of which the upper is greater. 


tin- most important. A conspicuous example is the well- 
known fn.nt of St. Peter's, at Rome (Fig. 83). 

In this the basement is suppressed, and the order below 
and attic above remain, constituting a division into two 

parts, the lower one much 
t he m< ist prominent. Many 
other examples might be 
added of a similar inequal- 
ity. In fact, if we go be- 
yond the limitation to the 
vertical wall surface, and 
take the roof into consider- 
ation, it is at once seen 
that every building with a 
visible pitched roof is there- 
by naturally divided into 
two parts, the wall below 
and the roof above, and 
that of these, the roof is 
usually much less in height than the walls. If these are 
taken int< > account, the number of buildings that are divided 
horizontally into two parts of which the lower is the great- 
est far exceeds that of any other type. 

There is another very frequent method of horizontal 
division, that, namely, into three parts; and this, like the 
double division, may be distinguished as of equal or of un- 
equal parts. 

Of three equal parts the Strozzi palace (Fig. 84) is an 
admirable example. 

Not less excellent as an example is the Vendramini pal- 
ace (Fig. 11), and the familiar Farnese palace is a third. 

In buildings that comprise pavilions as parts of their 
composition, this triple division is often combined with the 

Fig. 83. 
St. Peter's Church, Rome. 

Horizontal division into two unequal 
parts, of which the lower is greater. 



double division before described, the triple division being 
used for the pavilions, the double for the connecting or at- 
tached parts, Fig. 52, c. 

Of buildings that are divided into three unequal parts, 
the usual and normal type, so to speak, is that in which the 
middle part is considerably the largest, of which Fig. 85 
is a fair sample. This is analogous to the division of the col- 
umn into base, shaft, and capital, or of the order into ped- 
estal, column, and entablature, in either case marking the 
middle division as that which should predominate. 

The validity of this treatment, of dividing the wall 
into basement, shaft, and 
frieze, has received an in- 
teresting illustration of re- 
cent years in the many 
very tall buildings that 
have been erected. 

Lacking precedent, 
and thrown back upon 
the spontaneous aesthetic 
sense for guidance ; forced, 
too, to its exercise by the 
conspicuous size of these 
great buildings which does 
not permit their appear- 
ance to be slurred over 
or ignored, both architect 
and layman have united in approving the basement, shaft, 
and frieze treatment as the most pleasing and satisfactory. 

Briefly recounting our classification, we have found three 
classes of buildings with respect to their division by hori- 
zontal lines: first, those that are not divided at all, but form 
a single part, included between base and cornice; second, 

Fig. 84. 
Strozzi Palace. 

Horizontal subdivision into three e< [ual 


•• '•■'■«. 



i ZssQ 

rt ix.'tt a. [ ,i ; . 




Fig- 85. 
tl division into three parts, of which the middle part is greatest. 

1 ' 1 



One part 

Two eq11a.lj3a.rts 

Three equal juart.v 

those that are divided into two equal or two unequal parts ; 
third, those that are divided into three equal or unequal 
parts. These are shown in diagram in Fig. 86, and these 
are all of the possible 
variations of horizon- 
tal division in which 
the parts are one or 
two ; and three out of 
seven possible varia- 
tions where the parts 
are three ; and these 
two are all that are 
available in composi- 
tion. The four remain- 
ing three-part varia- 
tions can be used only 
by such treatment as 
will throw them into 
one or the other of 
these classes. 

Thus at a (Fig. 87), 
a six-storied building 
has three stories in- 
cluded in the upper tier of arched openings, two in the 
basement and one in the space between. Now there are 
two ways of regarding this middle space, either as a frieze 
to the basement, and in that case the upper one of the 
two mouldings which bound it must be larger than the 
lower, as at b, or as a base to the upper part, making the 
lower moulding the largest, as at c. Indeed, the smaller 
moulding might be omitted entirely, which would render the 
assignment of the middle part, whether to the upper or 
lower division, more clearly discernible. 

' ' | I' ' S 

I =1 I 

TWo unequal jjarti: Three unequal parts 

Fig. 86. 

Diagram of all possible modes of horizontal 


Fig. 88 is a diagram of the seven possible three-part 
arrangements comprising the original type of three equal 
parts and these of three unequal parts. 

Of these seven Nos. i and 6, together with the upper 
unnumbered figure, in which the parts are equal, are the 
types that have been included in the former diagram at 
Fig. 87. The rest can be treated only by a process such as 
we have described, by regarding them as compositions of 

r\ o 



f|A An 

r^ r\ r\ 

r^\ r^ r^ 

/*-\/-^ s~\/^\ ^> f~\ 

r^ r^\ r\ 

Fig. 87. 

Horizontal division into two parts, and assignment of small middle por- 
tion to one or the other. 

two parts, one or the other of which is itself a minor com- 
position of two parts. Thus in No. 3 the lowest part is 
naturally regarded as a base for the lower part of a two-part 
composition, while in No. 4 it must be regarded as a frieze 
tor tlic upper part. 

In the same way, if four or more parts are used, they 
must be subordinated to each other in such a way as to 
secure a clearly marked general division of the whole into 
two or three parts as we have indicated. 

II1 1 conclusion is, that if we exceed two or three parts 



Three e<jua.l purls . 

in our horizontal division, we can make the results pleasing 
only by subordinating the additional parts to the two or 
three that we have admitted as our foundation. If we fail 
to do this the 
whole composi- 
tion becomes an 
jumble of parts 
in which the eye 
discerns no fund- 
amental unity of 

But, with 
such subordina- 
tion, we can han- 
dle any reason- 
able number of 
minor parts, pro- 
vided that they fall easily into the simple groups of one, 
two, or three main parts that we have enumerated. 

There are notable analogies, it will be remarked, between 
the vertical divisions before discussed and these horizontal 

In the first place, the analogy in number is noteworthy: 
one, two, or three parts horizontally, correspond precisely 
to the one, two, or three masses which, as we have seen, 
may be used as vertical divisions. In the second place, there 
is a close analogy between the two equal or two unequal 
horizontal subdivisions and the two symmetrical or two 
asymmetrical masses of vertical arrangement. In the third 
place, the rule that in triple horizontal subdivision, the mid- 
dle part must be the largest, is paralleled by the fact that 
the central one of three masses must be the largest ; while 

* *• a 4- 5 6 

Three unequal parts- 

Fig. 88. 

Diagram of all possible modes of horizontal division, 
of which only three are available. 



three equal horizontal subdivisions match the like number 
in secondary, if not in primary masses. 

And the reason is the same as in the case of masses: the 
mind accepts one thing with one more attached, either on 
one side i >r on both sides, as a unit, but beyond this the sense 
of unity is lost, and a sense of multiplicity takes its place. 

So true is it that three is the limit of horizontal parts, 
that it is almost as difficult to find examples of four evenly 
divided parts as of the corresponding four-part division of 
vertical masses. Even where they occur, they show at- 
tempts to subordinate one part or the other, as we have seen 
should be done. 

Here are two examples (Figs. 89 and 90). In Fig. 89 
it is evident that there are four parts in the front. Not 

only are there four 
stories, but each of 
these is carefully 
separated from the 
next by a deliber- 
ate moulding. It 
is true that the 
stories are disting- 
uished, to some ex- 
tent, by a differ- 
ence in treatment, 
but this has not 
been carried far 
enough to over- 
come the first impression that there are four, or at least an 
impression of multiplicity instead of that of unity. The 
id story is manifestly superfluous; and it is easy to 
imagine how much the appearance would be improved could 
this story be omitted. 

Fig. 89. 
Academy of Fixe Arts, Vienna. 

Horizontal division into four parts, which 
would be improved by merging the second story 
into the first more completely. 



It is plain, too, that the architect has held this opinion, 
for he has done much to merge the second story in the 
first by similarity of wall treatment and by diminishing the 
moulding between these stories to a string, while elsewhere 
there are full cor- 
nices. Had this been 
carried still further, 
and this moulding 
omitted altogether, 
the design would 
have been much im- 

Another step 
might have been 
taken and the archi- 
traves of the second 
story windows omit- 
ted, reducing them 
to mere holes in the 
wall, more fully ob- 
taining, what the architect has evidently attempted, the 
complete fusion of first and second stories in one, and a 
satisfactory three-part division of the whole. 

Fig. 90 is another four-part division, with the usual 
scattering and unconcentrated effect. In this, however, it 
is the fourth story which the designer has tried to suppress 
with faint success. As it is, he has endeavored to make the 
whole story into a sort of cornice or frieze by a continuous 
series of vertical enrichments upon the wall surface be- 
tween the windows. 

If he had had the courage of his convictions, and had 
brought the overhang of his main cornice down to the very 
heads of the fourth-story windows, converted the cornice 

Fig. 90. 
Palais Epstein, Vienna. 

Division into four parts horizontally, which 
would be improved if the fourth story were 
treated more completely as a cornice. 


w them into a proper taenia moulding, and made the 
vertical objects into something more like the usual console 
or 1 Taekel. he might easily have completed the cornice ef- 
fect of the whole story, and have thus obtained the three- 
part effect for which he was striving. 

There are many other compositions in which, guided by 
instinctive taste, the architect has made an effort to reduce 
f<»ur or more natural divisions to two or three, but has 
stopped short of the desired end from lack of boldness in 
reaching out for a result which was not clearly formulated 
in his mind. 

Where the designer fully realizes what it is necessary to 
achieve, it is almost always practicable, by proper treat- 
ment, to fuse some stories into one, to nearly eliminate 
others, until we obtain a composition in a single part, or in 
two or three parts, any of which is sure to be satisfactory. 

Another very frequent use of apparently redundant 
horizontal mouldings is in the cases where a whole story is 
used as a band of separation, as often occurs in very tall 

It is not unusual, in these, above the four or five stories 
that form the basement, or below those that form the frieze, 
to find a story with mouldings above and below, and with 
enriched treatment of one sort or another, the whole mark- 
ing the line between the basement and superstructure, but 
n< it constituting a division of the wall surface, in any proper 
sense. This is seen in the enriched story below the upper 
arcade in Fig. 85. 

Before passing on to the consideration of other ways in 

which four or more horizontal divisions may be used, let us 

rt briefly to some of the analogies that we have before 

alluded to, between double and triple divisions, whether in 

L or horizontal sense. 


In double division, both vertical and horizontal, we can 
pass freely from the two equals to two unequals, either 
seeming quite natural and satisfactory. 

There is this difference, however, that in the case of two 
vertical parts, whatever be the inequality of size, similarity 
in general appearance is essential ; while in the case of hor- 
izontal division they may be either similar or dissimilar? 
certainly the latter when they are unequal, while there is 
little choice when both are equal, similarity and dissimilar- 
ity being both used with freedom. 

In triple horizontal division there is the same tendency 
for one part to be made larger than the others that we 
observed in the discussion of vertical masses. And there is 
the further analogy, that the largest part must be the mid- 
dle part. 

Apparent exceptions, such as the Doge's palace at Ven- 
ice, in which the uppermost part is the largest, and the mid- 
dle part the smallest, are often found to be, as in that build- 
ing, instances of a double division with one of these 
subdivided again into two. In this palace the two lower 
stories are of closely assimilated arched treatment, thrown 
together into one by that similarity, and also distinguished 
thereby from the upper part, which is a particularly flat, 
unbroken wall surface. 

There is a strong tendency, in many buildings, to this 
method of treating two of the wall surfaces alike, and the 
third in marked contrast. Such is the Vendramini palace 
(Fig. 11), in which the second and third stories are alike 
arched and columned, the first story quite different, being 
as far as possible a plain wall surface. 

Here, again, is a striking analogy with the tendency in 
groups of three masses, to make two alike and the third 
of an entirely different character, with such minor differ- 


ences only as are really unavoidable on account of the 
radical differences in the constructive relations of things 
piled on top of each other and things standing alongside of 
each other. 

The last way in which buildings can be divided horizon- 
tally presents an equally striking analogy to the correspond- 
ing arrangement of vertical parts. It is the continuous ar- 
rangement to which we here allude. 

'Just in the same way that four or more secondary parts 
in vertical grouping form a series that depends upon its 
continuity and uniformity for its unity; so four or more 
parts may be used in horizontal division under like condi- 
i i< >ns. 

Inasmuch as the parts are usually of necessity made to 
correspond to the stories, more than four parts means a 
building of considerable height, often a veritable tower; 
and accordingly we find a vertically continuous composition 
in such buildings only. 

In many towers, and in many of the recent tower-like 
buildings, which -ire arranged with a basement, shaft, and 
capital, we find that the shaft, instead of tall, uninterrupted 
piers, such as had the but lately fallen Campanile at Ven- 
for an example, has the whole middle portion cut up 
by horizontal mouldings at each story, into a series of slices, 
piled one on top of the other. 

In Fig. 91 are shown several of such buildings. At a is 
the Campanile of St. Mark's, with vertical treatment of the 
middle portion, for comparison with the opposite method, 
which is shown in the three other examples. At b is the 
Campanile at Lucca, in which the continuity of the middle 
stories is seen, all alike above the basement until we come 
to tlie topmost; at c is a pagoda, where a precisely similar 
arrangement is seen, although in a totally different style; 



while .it d is a modern office building. The fundamental 
identity of motive in b, c, and d is easily seen; each con- 
sisting of a top part and bottom part, with the larger middle 
part cut up into a series of uniform slices. 

Now the conditions which apply to this vertical contin- 
uous arrangement are the same as those for a colonnade, 
or any horizontal series: the units composing it must be all 
alike, and all of the same size. Even when a gradual dim- 
inution of size in the building requires a differenee in the 
diameter of the stories, the height of the stories must be 
equal, or only diminished very slightly to match the taper 
of the tower, so that to the eye they still appear to be equal. 
The design of the units, too, must be alike; the windows, 
if such occur, equal and arranged in the same way; other- 
wise the same s< >rt of hiatus would be felt as in the colonnade 
with a column missing. Variety may be all very well in 
its place, but a continuous series, vertical or horizontal, is 
distinctly not the place for it. 

At the most, perhaps a small balcony or some similar 
evidently extraneous embellishment might be introduced, 
corresponding to the central motive of the Farnese palace 
(Fig. 1 08), offering a point of concentration for the whole 
composition, while not interrupting the evenness of the 

A word may be said, in conclusion, as to the treatment 
of roofs. 

Roofs, as before noted, in buildings having visible roofs, 
constitute a natural division into two parts of the entire 
composition, the lower one being the perpendicular wall, 
the upper, the roof itself, of whatever character. 

The roof may be treated as the crowning member of a 
two- or three-part composition, the wall below being un- 
divided or divided into basement and shaft only; or it may 


receive separate treatment, as the upper member of a two- 
part composition, the wall surface being treated independ- 

Roofs naturally fall into two classes as far as composition 
is concerned: those that run up to a point and those that 
have a horizontal ridge. Many roofs will have one classifica- 
tion as regards the broadside, and another from the gable 
point of view. 

As to the pointed form, it makes little difference whether 
the point is that of a gable or that of a pyramidal or conical 
roof. The principle of treatment is the same in all. 

In the case of mansard roofs, or what are usually called 
mansard roofs, — the true mansard is more nearly what is 
commonly known as a gambrel, or rather it might be called 
a hipped gambrel, — in these the horizontal line is seen 
from every point of view, the upper part of the roof being 
either flat, or of so low a pitch as to be out of sight. 

The pyramidal outline, whether of gable or receding 
roof, is associated with the individual, perpendicular parts 
of the composition ; the horizontal roofs with the horizontal 
and continuous parts. 

Even where a building is a horizontal composition as a 
whole, topped by a comparatively low and long straight- 
ridged, hipped roof, the effect of the inclined lines of the 
roof in profile often is of the happiest in adding a sense of 
individuality to the building, without diminishing the gen- 
eral horizontality of treatment. 

The same advantage, in a converse sense, attaches to 
the use of a small ridge at the top of the tall roof of a pa- 
vilion. So used the truncated pyramid falls more easily 
into harmony with the horizontal lines of the rest of the 
roof than if its individuality had been still further intensi- 
fied by carrying the pyramid to an apex. 


As for horizontal divisions obtained by other methods, 
such as by overhanging or setting back, in one or more 
offsets, the whole upper part of the building, little needs be 
said. So rarely is such construction practicable, and so 
closely will the general principles that we have stated ap- 
ply, that a demonstration at length would be superfluous. 



THE word proportion is very often used as if good 
proportion, that is, pleasing proportion, were an 
inherent quality, quite independent of conditions or cir- 

We continually hear such phrases as "a well-propor- 
tioned window" and read discourses upon the ideal propor- 
tions of openings, and other things, as if proportion were 
something absolute — a desiderandum to be earnestly longed 
for and set before us as a goal, even though rarely attain- 
able — like virtue, an ideal to be worshipped rather than 

Thus some have laid down that in a properly propor- 
tioned window opening the height should be just twice the 
width; others have held that the so-called "golden section" 
was the proper thing, the width to the height being as the 
height to the sum of the width and height, a rule which 
gives incommensurable dimensions, approximately, as 5 to 
8, 5 being to 8 nearly as 8 is to 13, 8 times 8 giving 64, and 
5 times 13, 65. 

Now this view may be well enough in connection with 
the style of architecture for which it was devised. It is not 
sufficient for a system which aims at laying down general 
rules, applicable to all styles past, present, and future. 

The truth is that the proper relative dimensions for any 
part of a building, that is to say, the ratio between the 
width and height, cannot be definitely laid down without 



reference to the relative dimensions of the other parts, and 
especially to the relative dimensions of the building itself. 

The rule above alluded to for window openings will be 
found t" work well enough where the dimensions of the 
fr< »nt are in s< »mewhat the same ratio, the height a half or five 
eighths of the length, or even with wider variations; but if 
we were to lay out the windows of a Gothic church by this 
rule, the result would be strange indeed. 

A truer view we proceed to lay down, truer because gen- 
erally applicable; and found to give a clew to results that 
satisfy the aesthetic sense, beyond which there is no appeal. 

Proportion is very much the same in architecture as in 
arithmetic. It cannot be asserted of two quantities that 
they are in proportion. It is necessary that there should 
be two pairs of quantities, each pair having the same rela- 
tion between its component quantities. Thus in the rela- 
tion above given between building and window opening, we 
should have 

50 ft. 

25 ft. 


10 ft. 

5 ft. 

ength of 

Height of 

Height of 

Width of 


is to 




is to 


Thus, again, in Fig. 92 the rectangles at a are similar, 
base being to base as altitude is to altitude. 


while those at b are dissimilar ; the relations of base to base 
being different from those between the altitudes, the pro- 

6 : 1 : : 10 : 8 
iv »t being true. 

Broadly interpreted, this means that a building must 

nade up, in great part, of rectangles that are substan- 

ially the same in the relation of their linear dimensions; 



that is to say, our composition must be either of long and 
narrow parts or broad and short parts. 

Thus in Fig. 93 we have two elongated rectangles at a, 

'Kr e 


if. ff >| 1^4 


Fig. 92. 
a. Similar, and b dissimilar rectangles. 

and two more nearly approaching a square at b. Neither pair 
is exactly similar, but the approach to similarity is evident. 
When the similarity is perfect, as in Fig. 94, the diag- 
onals, as shown in 
dotted lines, will 
be parallel to each 
other. This gives 
a simple construc- 
tion for laying out 
a drawing that is 
often available. 
When the relations 
are inverse, that is, 
when one rectangle 




k-3~ * 





- 1 

Fig- 93- 
Two pairs of approximately similar rectangles. 

is horizontal and the other vertical, the diagonals, instead 
of being parallel, become perpendicular to each other, as 
shown at Fig. 95. 


— <c 

2 < 

It is not necessary in a composition that all of the rec- 
tangles, whether they arc long and narrow, or broad and 
sh«»'rt, should stand in the same relation to the ground line. 

On the contrary, 
some may be placed 
on their shorter 
sides, and some on 
their longer sides as 
1 >ases, without detri- 
ment to the result. 

Through these re- 
lations a composi- 
tion may be rapidly 
laid out, or the rela- 
tions of parts in an existing work maybe easily and quickly 
recognized. The diagonals are to be drawn approximately 
to tlie an-les of the parallelograms formed by mouldings 
and by vertical breaks in plan. Such measurements can be 

; 3-i 

Fig. m(. 
Two pairs of exactly similar rectangles. 

W <?■ 


- 6- 




Fig- 95- 
Two pairs of rectangles exactly similar and inversely proportional. 

approximate only, and this for two reasons, and without at 
.ill impugning the correctness of the proposition by admit- 
ting the impossibility of its precision. 

The first reason is that all that the eye requires is a rea- 



sonable approach to exactness. It is impossible to distin- 
guish without measurement whether such relations are ex- 
act or not. 

The second reason is that as the lines upon a front are 
marked by shadows of greater or less breadth, it is as much 
out of the question to insist upon precision of measurements 
as it would be in a charcoal sketch. There is no precise 
point to which to measure. 

A third reason might be added — namely, that as in mu- 
sic the slight departures from purity of tone caused by the 

Fig. 96. 
Disproportionate gables. 

harmonics that intermingle, so far from displeasing, give 
to each instrument the character of tone that distinguishes 
it; so in linear composition, trifling discrepancies give zest 
and prevent satiety. 

Viewed in this way, proportion is seen to be closely con- 
nected with the general similarity in shape of like parts of a 
building, that we have already found to be conducive to a 
pleasing result. 

If we place two gables, of different pitch, alongside of 
each other upon the same roof (Fig. 96), the result is any- 
thing but pleasing. 

The disagreeable appearance is due to the fact that the 

I 1-' 


gables are dissimilar, as is evidenl at a glance, one being an 
acute, and the other an obtuse triangle. 

If we reduce this to figures we shall have something of 
this si irt : 




Base of 



In I. 1 

df acute 



is to 

of obtuse 





a result which is seen to be untrue. 
Nor is it improved by inversion, 






Base of 


is to 

of acute 


of obtuse 

is to 






In either way the lack of similarity in the geometrical 
relations hears out the general lack of similarity in appear- 
ance which the eye attests and resents. 

In the last analysis these two conditions are one, for, 
however complicated the figures that we may compare, if 

their dimensions are geo- 
metrically similar, the simi- 
larity of their outlines will 
be at once recognized. 

In Fig. 97 are two arch 
mouldings of reversed cur- 
vature, enriched with crock- 
ets and finials. These are, 
in ordinary language, alike, 
except that one is about 
half the size of the other; 
that is to say, they are gen- 
similar in appearance, and any measurements that 
may be taken will be found geometrically similar. 

Fig. 97. 
proportionate arch mouldings. 














This law of equality of proportional relations, applied 
to parts of the first order, means that the primary masses 
with their links and appendages shall have the same ratio 
of height to width, or, failing that, an inverse ratio. This 
is shown diagrammatically in Fig. 98. 

Three typical cases are shown. In the first (a), all of 
the parallelograms are placed vertically; in the second (b), 
all are horizontal, and in both 
of these cases the diagonals 
are all parallel to each other ; 
in the third (c), some of the 
parallelograms are vertical 
and some horizontal, making 
the proportions inverse, and 
the resulting diagonals per- 
pendicular to each other. At 
Fig. 99 is an example of what 
has been said. 

In this the height of the 
main body of the building, 
from ground line to top of 
cornice, is almost exactly the 
same as its width ; that is, the 
front closely approximates a 
square; each of the appen- 
dages is also nearly a square, and the diagonal lines which 
might be drawn would therefore be parallel. The primary 
parts are therefore proportionate. 

Observe, too, that the roofs are of the same general shape. 
It is true that the main roof is somewhat higher in relation 
to its length, but all alike are long and low rectangles, ap- 
proximately proportionate to each other. A second series 
of proportionate rectangles exists in the door and windows, 







Fig. 98. 

Proportionate relations between 
parts of the first order. 



all of which are vertically elongated rectangles, even to the 
not numerically proportionate, notably so the second- 
story windows of the main building, yet all of similar char- 
acter, and some of them nearly proportionate. 

It is not necessary that every part of a composition 
should on form to the same ratio: there may be two or more 
sets of ratios, and certain parts that conform to each. 

Fig. 99. 

Townsend House, Washington, D. C. 

Excellent example of correct proportion between mass and appendages. 

Tims, in this instance, the mass and appendages are propor- 
tionate in one ratio, the roofs in a second, and the windows 
in still a third. 

An example of classic reputation, the Villa Medici, is 
shown in Fig. 100. Here, again, the mass and appendages are 
squares, as in the previous case. And not only this, but the 
square is carried throughout into many details. The inter- 
c< ilumniatit >ns of the loggia strongly suggest squares, although 
they are not squares. The four panels on each side of the 



Fig. ioo. 
Villa Medici. 

Mass and appendages are proportionate, being 
approximately squares. 

arch over the en- 
trance are squares, 
and so are those be- 
tween the windows 
above, those below 
the third-story win- 
dows of the appen- 
dages, and those up- 
on the anomalous 
pavilions or towers 
that stand upon the 
appendages. A sec- 
ond approximate 
proportion exists 
among the various 
rectangular win- 
dows, and the horizontal panels of the appendages; and 
still a third among the arched openings and niches. 

Still another classic Renaissance example is at Fig. 101, 

the Villa Bor- 
ghese. In this 
the ruling pro- 
portion is based 
on an approxi- 
mate ratio of 2 
to 3. 

The two pri- 
mary masses of 
which the front 
is composed are 
Fio-. IO i. proportionate to 

Villa Borghese. the arcaded link 

Proportionate relations of masses and link. wnicn Unites 

I 46 


them, as shown by the diagonal lines. Moreover, the height 
of the link is determined by the intersection of the diagonal 
with the inner vertical line of the mass, so that the upper 
story < >f the wing is necessarily in the same proportion. The 
lower story of the wing is also cut off in a like ratio by the cor- 
nier of the loggia, which is carried through ; while the row of 
horizontal oblong panels over the arcade, and all of the win- 
dow openings approach the same general proportion. Even 
the curvilinear niches, which usually would be circles, and 

which are circles in the 
previous example, are here 
el 'ii gated into ovals, in har- 
mony w r ith the oblongs of 
which the whole front is 

At Fig. 102 is a building 
of entirely different style 
and period, the Cathedral 
of Notre Dame at Paris, yet 
here again parallel propor- 
tionate relations prevail. 
Each of the masses is pro- 
portionate to the link, as 
shown by the parallel diag- 
onals, but in this case di- 
rectly instead of inversely 
as in Fig. 10 1, that is, both 
masses and link stand on 
their lesser sides as bases, 
while in Fig. 101 the base 
■ mass is the least side, that of the link is the greater. 
The height of the part of the mass that extends above 
e link is here determined by the intersection of the diag- 

i'i". 102. 
Notre Dame, Paris. 

Proportionate relations of masses and 



onal of the whole front with the inner vertical line of the 
mass, which is analogous to the relations shown in Fig. 
1 o 1 , although not precisely 

At Fig. 103 the propor- 
tional relations in one par- 
ticular are just the same as 
in Fig. 102. 

In both of them the 
main diagonal shows that 
the ratios of base to height 
of the part of each primary 
mass above the link are 
equal to those of the lower 
part of the primary mass 
plus the link, and also to 
the ratio of base to height 
of the whole front. l 

In Fig. 102, however, 
the relations between link 
and mass are quite differ- 
ent from those that obtain 
in Fig. 103. In the former, 
these relations are the typi- 
cal ones shown in diagram (Fig. 98, a) of direct proportion 
between mass and link ; while in the latter no relations of 
similarity are traceable between the corresponding parts, 
much to the detriment of the appearance. 

This lack of proportion has the effect of making the link 
subordinate in height, indeed, but superior in width, as com- 
pared with the two primary masses, resulting in an uncer- 

1 In Fig. 103 the diagonal is not drawn, but the edge of a card laid 
over the figure will show the relations clearly. 

Fig. 103. 
Cathedral, Marseilles. 

Disproportion between masses and 


taint v as tn which is meani to be subordinate to the other, 

and this is one reason why the modern design is inferior to 

the mediaeval. 

Volumes might be occupied with the comparison and 

analysis of such proportional relations in various buildings. 

It is our aim here merely to suggest the method, leaving 

it to the student to make the application. 

It will be found that 
the examination of the 
buildings of the past and 
of the present acquires an 
entirely new interest when 
studied in the light of the 
laws of composition that 
we have adduced. Scarce- 
ly one will be found to 
conform to them all, but 
there are still fewer, and 
none of the best, that fail 
to conform to some of 

Figs. 104 and 105 are 
examples of failure in 
maintaining proper pro- 

In both of these the 
fault is the same — the tall, 
narrow, corner turrets have 
no relation of similarity 

with the main bulk of the building. It is broad and 

square; they are lean and thin; they are in ordinary 

phrase quite "out of keeping" with the primary mass of 

the building. 

l"ig. 104. 
Church ok St. Nicholas, Potsdam. 

Lack of proportion between the 
broad primary mass, and the two tall, 
narrow secondary masses. 



It is often a cause of wonder why the tower of the Flor- 
entine Palazzo Vecchio stands as it does, cut short, upon an 

impending parapet. One 
reason, at least, is, that to 
have continued it to the 
ground would have given 
this same unpleasant con- 

Fig. 105. 

Synagogue, Berlin. 

The same fault as that shown in 
Fig. 104. 

trast between the massive - 
ness of the building and the 
attenuation of the tower 
(Fig. 106). 

Another instance of a 
secondary mass that con- 
forms to the law is shown in 

Fig. 107, where the ratio of the height to the width of the 
portico is equal to the same ratio in the whole front, pro- 

Fig. 106. 

Palazzo Vecchio, Florence. 

The tower, if continued to the 
ground, would seem disproportion- 
ately tall and narrow. 


during a particularly happy effect, which is felt before the 
cause of it has been assigned. 

In the house shown in Fig. 60, the >r ginal intention was 

.rrv the oriel windows down to the g <und, forming bays 

in each story. Ii was found that this gavs two tall, narrow 

Fig. 107. 

Public School, New York. 

The portico is proportionate to the whole front. 

objects, quite at variance with the horizontal motive that 
had been adopted as the keynote. 

They were therefore shortened until the width was some- 
what greater than the height, to correspond with the general 
dimensions of the front, although precise numerical propor- 
tion could not be attained. The rectangle of the veranda 
also preserves the same preponderance of width in com- 
parison with the height. 



The same law extends to the details of a building, of 
which the most conspicuous, in modern buildings, are the 
windows and doors. 

In some of the monuments of the past the numerical 
relations of the details are very easily traced. The Farnese 
Palace, for example (Fig. 108), shows approximately these 
relations from measurements of a larger drawing. 1 


Whole front 136 

Main door. 

2d story window, clear opening. . 
2d story window, clear opening. . 

3d story window, clear opening . 
3d story window, clear opening . 



Appromm \ n. 

28 to crown 
of arch. 

24 including 12 outside 
pedestal columns, 
and archi- 
traves, but 

not pediment. 

25 including 13 outside 
pedestal and columns, 
architrave, but 

not pediment. 

These dimensions show everywhere an almost exact 
ratio of width to height of openings of i to 2, which is the 
inverse of those of the whole front. 

Just how far the architect had in mind such a close nu- 
merical relation we cannot tell, but the fact is conspicuous, 
that in striving for a pleasing result he did adopt it, whether 
consciously or unconsciously. 

1 These measurements, with those that are given of other buildings, 
are only relative, having been made from larger drawings and photographs 
by scale, and do not represent any actual dimensions in feet or any other 



In the Bartolini Palace (Fig. 109), the relations arc as 
follows : 


Approxima 1 k 

Height Width Ratio 

Whole front 160 120 4 : 3 

Each of the 3 stories into which the front is divided 1 : 2 

Main door 30 15 2:1 

Main door 40 to top of 29 outside 4 : 3 

architrave. columns. 

2d story window, clear opening. . 24 12 2:1 

2d story window, clear opening. . 3 1 to top of 21 outside 4 : 3 

architrave. pilasters. 

3d story window, clear opening. . 24 12 2:1 

3d story window 31 to top of 2 1 4 : 3 


Here we no longer find the same simple ratio prevailing 
in all parts. 

The ratio of the front, width to height, the height here 
being greater than the width, is 3:4, which is found again, 
approximately, in the outside dimensions of the main en- 
trance and of the windows in the upper stories. The three 
parts into which the front is divided are of slightly different 
height, although apparently equal, the relation of height 
to width of each being approximately that of the openings, 
1 : 2. 

Another example of a widely different type is the Hotel 
de Ville of Paris, of which the central portion is shown in 
Fig. no. 

The first column in the table contains the scale measure- 
ments. The second column contains the ratio of the meas- 
urements with considerable accuracy. The third column is 
the nearest simple ratio to which the actual ratios approx- 

Thus the first item, the ratio of the total height of the 

is \ 


vertical wall to its total length is nearly ,;,, approximating 
'. ; very nearly the same .is the ratio of the dimensions of the 
vertical walls of the two pavilions, taken inversely; which 

Fig. 109. 
Bartolini Palace. 

ortional relations subsist between the whole front and the various 


is \. Comparing these by reducing to a common denom- 
inator we find the corresponding ratios T 4 r 9 T and T 4 T 8 2-, a differ- 
that the eve cannot detect. 
'I he win. lows have the correct academic dimensions, the 
width being one-half the height, with the notable exception 



of the most conspicuous row of windows in the front, those 
in the central part of the " bel etage," which are three times 
their width in height. This again corresponds with the di- 
mensions of the facade below the cornice of the central link', 
and exclusive of the part of the pavilions that projects 
above jt. Of this the ratio is Trr , inversely, which compares 

with the preceding as yf- with 


4 s- 


Scale meas- 
ures — height 
to width 

Total height to cornice of pavilions ami 

total length of front I 28 

Width of pavilion at third story, and 

height to cornice of third story 28 

Roof of central link [6 

First-story window 3 

Second-story pavilion window 
Second-story window in central link. . . . 
Height to cornice over second story, and 

total length of facade ! 20 

Central portion of second story I 8 

Turret at each side j 16 

Fleche above ridge | 20 

Central link, height to cornice of second 

story and width between pavilions ... I 20 
Portion of pavilion above second-story 



Lower part of clock 


i 1 . 

3 2 


Actual ratios 




5 : 

3 : 


i : 2 

1 : 2 

1 : 2 

1 : 2 

1 : 2 

1 : 3 

2 : 3 

2 = 3 

2 : 3 

3 : 4 

Corresponding relations are found in the dimensions of 
the various parts of the Greek temples. The conspicuous 
fact in the proportions of a Doric temple is the continual 
recurrence of horizontal rectangles of differing sizes, but of 
closely related dimensions (Fig. in). 

Above there is a row of the butt ends of the mutules; 
next below a corresponding row of regulas, which are of 
equal dimensions with the mutule ends in most cases. Be- 



low these, again, is the row of llat faces of the abaci, also 
elongated horizontal rectangles. All of these are closely re- 
lated among them- 
selves, and also to 
the vertical rectan- 
gles of the columns. 
Taking the dia- 
meter at the base of 
the column, the re- 
lation of width to 
height is substan- 
tially the same as 
that of height to 
width of the abacus, 
a] (proximately as 2 
is to 11. 

In one instance, 
at least, that of 
the great temple at 
Olympia, both the regula and mutule-end exhibit the same 
relation, but in most cases they are but half as broad in 
proportion to their length, having a relation of about 1 to 
11. The fillets of the triglyphs exhibit the same relation 
very nearly, being as 1 to 12, while the channels between 
them are as 2 to 12, approximating the first ratio of 2:11, 
which is exactly repeated in the caps of the triglyphs. 
These results may be tabulated as follows: 

Column, diameter at base to height 2 

Abacus, height to width 2 

Cap of triglyph, height to width 2 

Channel of triglyph, width to height 2:12 

Fillet of triglyph, width to height 1:12 

Mutule end, height to width 1 

Regula end, height to width 1 

Fig. no. 
Hotel de Ville, Paris. 

Proportional relations subsist between the 
various parts. 



showing the close conformity of all parts, large and small, 
to one or the other of two simple ratios. 

These are the really important relations, of which the 
eye takes immediate notice, such relations as those between 
the width of the front and the height to the apex of the 
pediment being of little 
practical avail. 

Nor are the dimensions 
in plan, nor the interior di- 
mensions of any building of 
any meaning at all in con- 
nection with this part of our 
subject; proportion being 
essentially the relations of 
objects that are seen at the 
same time, and not of those 
which are only known by 
a comparison of measure- 

The following rules may 
be laid down for practical 
work, and with these many 
compositions of recognized 
merit seem to agree : 

First. Let the front of 
the building and its various 
parts all conform to the 
same general type of rect- 
angle, either the notably 
broad or the notably nar- 

Second. Let parts that from their difference in func- 
tion are unavoidably different in dimensions, as columns, 

Fig. in. 

Proportional relations between the 
various parts of the Doric order. 


Fig. i i j. 

rooi ;b tuld be pro- 
portionately related to the 
building below. 


for instance, differ from doorways, 
conform to separate ratios. 

Third. Instead of exact simple 
ratios, let the nearest convenient 
approximations be used, thus in- 
troducing slight differences into the 
general similarity. Instead of mak- 
ing a window opening precisely twice 
its width in height, make it n : 23 
or 5 : 1 1 , or 6 : 1 3 , and use different 
variations in different tiers. 

There is one application of the 
principle of similar dimensions of 
which we must speak before con- 
cluding, that is, to the proportion- 
ing of roofs. 

In a general way it may be said 
that the roof, whether gabled, pedi- 
mented, or hipped, should conform 
to the mass below. If the mass is 
tall and narrow the roof should be 
tall and narrow; if the mass is low 
and broad, the roof should be low 
and broad. 

Thus in Fig. 112 the broad and 
comparatively low mass of the Greek 
temple at a has a pediment of con- 
formably low pitch, which would be 
included in a proportionately long 
and low rectangle. At b the Roman 
temple, with its loftier order, has 
one somewhat higher. Romanesque 
buildings of moderate proportions 


J 59 

have still higher roofs, as at c, while taller buildings of the 
Gothic period have the roof still further elongated upward 
to match, as at d, culminating in the tower with the spire as 
its appropriate capping, as at e. 

Now this upward stretching of the triangle of the roof 
as the rectangle of the front elongates is not merely a mat- 
ter of historical interest, because roofs thus disposed are al- 
ways appropriate in relation to the fronts upon which they 
are placed, whatever be the details of style adopted. As a 
matter of fact this principle of assimilating the roof to the 
mass below is instinctively carried out in the various parts 
of a complex front at all periods, and at the present time. 

For illustration, glance again at the Hotel de Ville (Fig. 
no), and note how the pitch is steeper and the relative 
height greater of the tur- 
ret roofs than of those of 
the pavilions. 

Curiously stumpy the 
turret roofs would look if 
they were made of the 
same pitch as the pavilion 
roofs, in spite of the gen- 
eral dictum that all the 
roofs of a building should 
have about the same pitch. 

Nevertheless here, al- 
though the pitch is really 
so different, the effect is of 
about the same pitch, as 

the reader may find if he will take the trouble to lay out a 
corresponding tower and tourelle with roofs of equal pitch. 
The result will be somewhat like a in Fig. 113, where, alth< >ugh 
of parallel pitch, the tourelle roof looks too low, while in b, 

Fig. 113. 

Roofs of taller and narrower parts, 
as the attached turrets, should them- 
selves be taller and narrower relatively. 


where the pitch is relatively steeper, it lo< >ks about the same, 
until we examine it critically, when we see the difference. 
Often the pitch of the roof , whether gabled or pyramidal, 

Fig. 114. 
St. Paul's Cathedral, London. 

is parallel toa diagonal of the whole front, or of some impor- 
tant horizontal division of the front. 

In St. Paul's, London (Fig. 114), the pitch of the central 
pediment is exactly parallel to the diagonal of the second 
order of columns, and very closely to that of the whole front. 

I'koi'oktu >n 

if. i 

In the Hotel de Ville the lines of the pavilion roofs arc 
approximately parallel to the diagonal of the front wall of 

the pavilion, as are those of the turret roofs to the diagonal 
of the turret above the corbelling. 

Going back to Fig. 35, c, the truncated pavilion roofs, 
again, have precisely the same pitch, being parallel to the 
diagonals of the masses below ; while at Fig. 115, the line of 
the main roof of each pavilion is inversely related to the 


iffrv -£&. -SSv -idfev 

• i^frTrfn 

j "J""!,! n 

Fig. 115- 
Design for a New York Residence. 

The peaked roofs of the pavilions are of equal height with the faces of 
their walls below, and might be inscribed within them. 

building, being perpendicular to the diagonal of the whole 
front below. The roofs of the pavilions are also of the same 
height as the walls of the pavilions from water table to cor- 
nice, and therefore parallel to the diagonal of half the wall. 
Note, also, that the roofs of the little dormers are made 
steeper than the main pavilion roofs, the faces of the dormers 
themselves being also more drawn out vertically than those 
of the pavilions to which the main roofs correspond. The 


same thing is noticeable in Fig. 62, where the tourelle roofs 
attenuated in comparison with the main roof. 

In classical buildings, on the other hand, in which hori- 
zontal treatment predominates, the general dimensions arc 
less in height than in length, and the pediments share this 
character to just the same extent. An example of a hexa- 
style Doric temple, at Fig. 116, shows that the rise of the 
pediment is equal to the height of the entablature, that is, 

the pitch is parallel to the 
diagonal of half the pedi- 
ment, being much the same 
sort of relation found in 
Fig. 115. This was not pre- 
served in the octastyle 
temples, the rise being 
greater than the entabla- 
ture, which raised the rela- 
tive, but kept the same ac- 
tual pitch. 

In the Ionic temples 
the rise of the pediment is 
somewhat greater relative- 
ly, as is the height of the order in relation to its length, but 
no such simple relation can be detected as prevails in the 
I )( wric. Still greater was the relative height in the Roman 
Corinthian, corresponding to a still further addition to that 
of the order itself, and to its elevation on lofty stylobates, as 
in the Pantheon, in which the line of the pediment is related 
to the whole order, being parallel to a diagonal of the order, 
in a way very precisely similar to that seen in Fig. 115. 

This is the artistic view of the phenomenon, of the grad- 
ual raising of the pitch of the roof from the Greek to the 
culmination of the Gothic period. 

Fig. 116. 
Temple <>f Neptune at Paestum. 

The pediment ise<iu,'il in height to 
the entablature, and might be in- 
scribed within it. 


Historically, this development is merely an interesting 
fact, set forth by historians in their dry-as-dust fashion, as 
they set forth other facts, without relation to cause prece- 
dent, or effect subsequent; while from a constructional 
point of view the steepening has been ascribed to the more 
inclement climates of the north that needed a higher pitch 
to shed the rain, and more especially the snow. Neither 
of these is a sufficient explanation. The historical view 
does not pretend to explain at all, and as for the construc- 
tional view there is too much to contradict it, most of all 
the fact that the rise of the pitch occurred in all climates. 
The phenomenon finds a rational explanation only in an ar- 
tistic one, the instinctive demand of the eye for a steeper 
roof on a relatively higher building. 

In the case of certain buildings, naturally vertical, that 
is, in which the vertical dimensions considerably exceed the 
horizontal, but which are treated horizontally, as is done in 
some Italian campaniles, the roof may be proportioned 
to the uppermost of the horizontal divisions taken by itself, 
as if there were no building below it ; or it may be propor- 
tioned to the whole building inversely, the lines being drawn, 
not parallel to the diagonals, but perpendicular to them, 
and including either the whole mass below or an important 

When it comes to determining the relative heights at 
which the horizontal mouldings should be placed, with which 
a building is usually subdivided, the general rule of propor- 
tion of similar rectangles does not apply. 

By the very nature of the case there can be no similarity 
between the rectangle of the whole front and the minor 
rectangles into which it is thus cut up ; nor can there be any 
similarity between these rectangles, unless, indeed, they are 


As a matter of fact they wry often are equal, or sub- 
stantially so ; not only in the classical styles, where the super- 
imposition of orders would lead naturally to an approximate 
equality, nor in storied buildings only, in which the practi- 
cal requirements would produce the same effect; but in 
styles distinctly non-classical and in buildings that are but 
one-storied halls of the usual church type. Both in two 
equal parts and in three equal parts there are many exam- 
ples of all styles and periods. 

There is some doubt in comparing measurements of 
heights in facades, as to the proper point to which the fig- 
ures should be taken, arising chiefly from the depth of the 
mouldings and of the friezes and bands which constitute the 
lines < »f demarcate >n. The stronger the lines, and the bolder 
the challenge to our sense of just proportion, the more diffi- 
cult it is to test it by figures. Bear this in mind when testing 
tlie following statements; bear in mind, too, that substan- 
tial correct dimensions, measured to such leading lines as 
tlie eye marks at a glance, are all that is needed. 

Of those in two equal parts the Banqueting Hall at 
Whitehall (Fig. 80), is an admirable example. Not only 
are the first and second stories equal, but the balustrade 
above the second is equal to the stylobate below the first. 
Another is the Stoedel Art Gallery (Fig. 52, b), and still 
another is the Villa Giulia, of which no illustration is here 
given, as it is an approximation merely, the second story 
being six-sevenths of the first in height. 

There is a most interesting case of juxtaposition of new 
and old, Renaissance and mediaeval, showing at once the 
likenesses and differences between them, in the Library of 
Si. Mark and the Ducal Palace at Venice. The former is 
frankly divided into two equal stories; the latter appears at 
first as in three parts, as indeed it is, but upon measurement 


it is found that the sum of the lower two is equal to the 
third. The height of the lower half is so subdivided that, 
if the lowest arcade be called 4, that above it will be 6, 
and the wall above 10, making the proportionate relation 
which has before been noticed, 4:6 :: 6:10 nearly, strictly 
it should be 9, or rather all of the numbers should be ad- 
justed fractionally to make an exact relation. This relation 
is called the "golden section," and is such that the first of 
two quantities is to the second as the second is to their sum : 

a : b :: b : a + b 

Another instance is at Fig. 82, in which the basement 
is to the main story as 3 is to 5, as close an approximation 
as whole numbers can give : 3 : 5 :: 5 : 8. 

In addition to the Ducal Palace we may cite another 
example of a building which is divided into three parts, 
two of which are found to equal the third, although the 
building is separated from the former by leagues and 
centuries, the Vienna Academy of Fine Arts, Fig. 89. 

In the Ducal Palace it was the lower half that was sub- 
divided ; here it is the upper half. No longer now by the 
"golden section," the division being nearly equal, say as 
6 is to 7, although this is an excellent specimen of the 
difficulty that occurs in knowing just wdiat lines to choose 
for measurement of heights. The lowness of the arches in 
the topmost part makes that part seem lower than it really 
is, while the shadow of the overhanging cornice gives no 
precise limit for measurement. But there is substantially, 
as said, an equal division of the whole height, and an equal 
division of the upper one of these parts. 

The lower part, it is true, has a line of entresol windows, 
which appears at first as a fourth division, yet these are 
merged in the first story by the similarity of wall treat- 



ment, and the uni< m wi >uld have been more complete had the 
moulded course below the entresol been entirely omitted. 

Fig. 117. 
Cathedral at Pisa. 

The front is divided horizontally into three nearly equal parts, of which 
the upper two are again divided each into two equal parts. 

Passing on to unmistakable three-part division, we may 
instance the Lyons Bourse, Fig. 5 1 , where the division seems 



at a glance equal, but is found to be, first story, 6; second 
story, 6 ; third story, 4, a difference far greater than would 
be supposed. The wall of the Strozzi Palace, Fig. 84, below 
the cornice and frieze is cut up by stringcourses into three 
parts that are, beginning at the top, as 7, 8, and 8 ; a much 
closer approximation to equality. In the Farnese Palace, 
Fig. 108, still beginning at the top, they are as 13, 12, 11, 
which is still closer, but in reversed order, the highest being 

An earlier example is the Palazzo Vecchio, Fig. 106, in 
which the division is 4, 8, 8, suggesting that of the Lyons 
Bourse, above mentioned. 

Another, and a peculiarly valuable one, is the front of 
the Pisa Cathedral (Fig. 117). It is peculiarly valuable, 
first, because it is divided into stories entirely as a matter 

of beauty, and not at all for ^.. ___ s t ^ — e 

utility; secondly, because it is 
a building upon which were 
lavished all the study and skill 
that were possible; and in the 
third place, because style at 
that time was fluent and not 
fixed, and men were at liberty 
to do what seemed to them 
most beautiful, without regard 
to authority or precedent. 

Minor variations undoubt- 
edly occur and are essential to 
its charm, but the main skeleton 
of dimensions in its laying out 
was clearly a triple division, of which the large first story 
arcade was the first, with two tiers of the smaller arches to 
each of the others. 



Fig. 118. 

Diagram of heights of horizon- 
tal divisions of front of Notre 
Dame at Paris. 


xuju' nmu.aj^^ 

Another very curious instance of irregularity based upon 
regularity is in the facade of Notre Dame, of Paris, shown 
in outline in Fig. ei8. No relation is discernible, at first 
glance, in the heights of the rows of arcades that adorn it; 
upon measuring them we find the peculiar i, 2, 3, 4, 5, or 
rather, 5. 2, 3, 1, 4, dimensions shown. These leave us 
more than ever al sea, as there seems nothing but caprice 

in their arrangement, until 
we notice that the sum of 3 
and 2 is 5, and that the sum 
of 4 and 1 is also 5, making 
+he whole an exact, although 
not precise, division into three 
equal parts. 

It would seem from these 
instances that some of the 
most delicate and carefully 
studied height relations in 
the buildings of the past have 
been based upon a funda- 
mental division into equal 

Still another example is 
seen in Fig. 119, the ancient 
gateway of the city of Lubeck 
—which resists all attempts 
to reduce it to figures, until we hit upon the fact that it is 
divisible very closely into three equal parts, measuring at 
either extremity and disregarding the irregularity of base 
line caused by the hollowed-out roadway; and if we call the 
whole height 27, and each of the three main divisions 9, of 
these three the uppermost is quite naturally divisible into 4 
and 5 parts respectively. 


Fig. 119. 
Gateway, Lubeck. 

Divided into three equal parts hori- 


So that our first rule will be somewhat thus: Divide the 
front into either two or three equal parts, or such as shall 
seem equal at first glance. 

The second rule will be: Divide the building into two 
or three approximately equal parts, and then subdivide 
some of these, in simple numerical ratio, in such a way as to 
distract the attention from the original division into equal 



IT might seem that in speaking as fully as we have upon 
Similarity, we had by implication said all that needs 
to be said upon the subject of its converse, Contrast. Per- 
haps, however, a separate discussion of the latter may 
thr< »w s< ime further light upon it, and involve a closer atten- 
tion to its correlate, Similarity. 

Than these two principles there is none more habitually 
outraged by the carpenter-designer, who invariably ruins 
what might be passable, by sticking in "for variety" things 
that should not vary, balancing segmental pediments 
against pyramidal, putting peaks where horizontals are 
needed, breaking up the line of frontage in plan without 
rhyme or reason, and distracting what should be harmonized 
generally, under the plea that he is doing it "for variety." 

S( i that it becomes doubly important for us to generalize, 
and distinguish, if possible, the cases where we must use 
contrast, where we may or may not at choice, and where 
finally we are forbidden entirely to use any contrast at all. 

There are three ways in which contrast may exist, 
ignoring for the present the fourth way, contrast in color — 
which we are not to take into consideration at all. 

To lu-gin with there is contrast in shape, as of a triangle 
with a square, or a semihexagon with a semicircle. Next 
there is contrast in size, as of a portico 50 feet long and 
20 high, with the building behind it, 150 feet long and 60 
high. Last there is contrast of position, as when a narrow 

1 70 


I 7 l 

rectangle is placed perpendicular to another narrow rectangle, 
as column to epistyle ; or as when a tall turret is placed upon 
a long and low building. 

These three then, contrast of form, of size, of position. 

What we usually mean when we speak of contrast, what 
the carpenter means when he puts a round turret on one 
corner, and a square one on the other, is contrast of form. 

The first rule is that contrast of form in any part is 
not an essential quality. It is perfectly possible to dispense 
with contrast in form entirely; to make all the parts of a 
building, little or big, that serve similar purposes, more or 
less close reflections of each other in form. Such was the 
Gothic style in its most perfect development. Here every 
turret and buttress was crowned with its crocketted pin- 
nacle, exactly reproducing the general outline of the spires 
upon the central and western towers ; every vault, window, 
and door was spanned by the curved triangle — every roof 
likewise of triangular outline. 

In the classical style there is the same avoidance of 
contrast of form. Every part is a more or less elongated 

The only really noteworthy contrast is that of the 
triangle of the pediment, and even this is kept as flat and as 
near to the horizontal as possible. It is interesting to 
observe that a similar contrast in the typical Gothic church 
is that of the rose window over the entrance, often the only 
important circular form in the building; and used, precisely 
as the pediment was used, to give unity to the main front 
by the introduction of a single contrasting and very much 
individualized object. 

In more modest and more modern designs the same 
principle of similarity throughout, to the exclusion of any 
striking contrast, is shown in Fig. 60. All the roofs here 



are hipped, main roof, dormer, and piazza roofs; and, al- 
though practical and optical considerations forbade making 
them all ^i precisely equal pitch, or all of octagon plan, the 
general similarity of effect is maintained, and the central 
unifying detail, the triplet window of PaUadian suggestion 
in this ease, as in that of the classic and mediaeval types 
above described, is the only contrasting form in the front. 

In the same way a good design may always be made by 
ling all ] >arts of which the functions are the same, alike 
in shape, all r< >< >fs, for instance, of the same pitch, all windows 
square-headed or all arched, and, even when the functions 
differ, the similarity may be maintained, if at all compatible 
with the performance of their functions, without detriment 
tot lie a] >] learance ; as when a roof which should slope to best 
fulfil its function, is made flat to harmonize with a general 
flat, horizontal treatment. 

Observe particularly that the lack of contrast in form 
does not prevent the greatest degree of contrast in size, as 
between the spire and the pinnacles, nor of position, as 
between t lie vertical oblong of the column and the horizontal 
one of the epistyle. 

We have spoken of cases in which contrast of form may be 
omitted ; we must note briefly one case in which it must be 

This is among members of the same group, whether of 
primary or secondary masses, or of details. It should be 
unnecessary to speak of a fact that is so obvious, were it not 
that unskilled designers in their desire for "variety" often 
fall into this very mistake. 

A square-corniced pavilion at one end and a gabled one 
at the other is a favorite device of the carpenter-designer 
n'n his design in an abortive attempt to do something 
particularly striking. 



Between parts of the same order, that is to say, between 
primary or secondary masses and their links and appendages, 
there may or may not be contrast whether of form, of size, 
or of position; as at Fig. 120, in which the same composition 





f ' 

UUU4 ** 



1 ; ■ — ~* 


Fig. 120. 

Contrast of form between the masses and their link and appendages is 
shown at a, similarity at b. 

is shown with the links and appendages treated in contrast 
of form, at a, that is, with pitched roofs, while the twin 
masses of the group have no visible roofs, and at b in which 
neither the masses nor the links and appendages have any 
visible roofs, all being treated alike. Contrast of size is 
shown in both a and b, and between the masses and the 
links there is also contrast of position ; and the contrast is, 
as it should be, if possible, greater between the appendages 
and the masses than between the link and the masses. 

A like difference exists in a and b, Fig. 121, in which 

a b 

Fig. 121. 

Contrast of form between the masses and their link and appendages is 
shown at a, similarity at b. 



a shows contrast of form in treatment; b, similarity. In 
the first c.isc, a, the primary masses have fiat tops, while 
links and appendages are pitch-roofed; in b all are roofed 
alike Either arrangement may be made to look well. 

Fig. 12 2. 

The appendages at <; are similar in form to the mass; at b they are in 


It will n< >t do at all, however, to make one tower pitch-roofed 
and one flat, as this would introduce a contrast in the mem- 
bers of the group which never produces an effect otherwise 
than lamentable. In this case the contrast in size is the 
same as in the previous example; while the contrast in posi- 
tion of the link no longer prevails, both masses and link 

being placed similarly erect. 
The same rule prevails in 
the case of the single mass, 
with either one or two appen- 
dages, which may be either 
similar to the mass or in con- 
trast with it. 

Thus in a country-house 

design, it is practicable to 

make the appendages like the 

mass as at a, Fig. 122, all 

with pitched roofs and gables, or the former may be flat- 

d attachments of a high-roofed mass as was often done 

in colonial houses, or perhaps mere open verandas, which 

Fig. 123. 

The roofs of the mass and appen- 
are similar in form, but con- 
'1 in position. 



lend themselves far better to purposes of composition when 
placed as appendages than when extended along the whole 
front. Sometimes, as in Fig. 123, the mass and appendages 
may both have roofs alike in form, in so far as both are 
visible, pitched, hipped roofs, but contrasting in position, 
that of the mass high pitched , those of the appendages 1< >w 
pitched. In such cases it of- 
ten looks well if the roofs are 
made of complementary pitch, 
with slopes perpendicular to 
each other instead of parallel. 

In the case of three masses, 
contrast of form, if used at 
all, must accompany contrast 
of size. When the masses are 
of nearly the same size con- 
trast of form should not be 
used ; but as the relative size 
of the central mass is in- 
creased, it may be of more 
strongly contrasting form. 
The links and appendages are 
subject to the same rules as in 
the case of two masses. 

Between secondary masses 
and primary, contrast may 
often be used with very great 
gain in what is commonly 
called " snap." 

In a, Fig. 124, the dormers 
are in contrast of position 

with the main roof, the former being of very high pitch, the 
latter rather low. Here again the pitch may often be 

Fig. 124. 

a. Contrast of position, b. Con- 
trast of form, c. Contrast of both 
form and position between primary 

and secondary masses. 

i ;<> 


complementary with good effect. In b, the semicircular 
portico is in contrast "I" shape with the straight lines of the 
primary mass, and in c the two bays arc- in the same sort 
of contrast with their primary. In all eases contrast may 
prevail between secondary and primary masses if desired. 

In details, which in modern work are, for the most part, 
windows and doors, contrast, either of shape or size, is 






□ cd 






Fig. 125. 

formation of continuous treatment into a triple group of details by 

contrasting size. 

necessary when it is desired to pick out and group together 
certain individuals in double or triple combinations. 

A triple group is formed, as shown in Fig. 125, from a 
row of windows that might have been uniform, by enlarging 
some of them to make the desired group, leaving the rest 
small. The smaller ones in such a case play the part of 
links and appendages to the larger ones that constitute the 
group, and, as in the ease of these, may be either of similar 
•ntrasting shape. 

This principle is often used in many-storied buildings, 
some of the openings being thrown together under arches, 


which become the most conspicuous parts of the c< >mpositi< m 
and the effective components of the group. 

We have thus enumerated and classified the principal 
cases in which contrast is possible. It is not permissible, 
as we have seen, to make any and every part < >f a c< >mp< >siti< >n 
of all kinds of shapes and sizes simply for the sake of an 
alleged variety; on the contrary, the principle of contrast 
must be used to form, or to accentuate, groups of similar 
objects, whether of primary or secondary masses, or of 



TIM-! foregoing principles have been developed in very 
much the same order in which they will spontane- 
ously unfold themselves in the mind of a designer who gives 
some thought to the apparently instinctive processes by 
which he proceeds. 

Alth< tugh the external appearance has been treated apart 
from any suggestion of the plan, it is by no means intended 
to convey the idea that an exterior should be, or can be, 
independent of the interior arrangement. On the contrary, 
it must, as lias so often been pointed out, hang upon and 
spring from the organic internal disposition. 

The organic internal disposition, however, is somewhat 
different from the accidental, arbitrary, or unorganized 
internal disposition. 

Considered as an organism, a building is arranged 
internally in the same way as a vital organism, that is 
to say, upon lines of communication between the various 

Just as in the body there is an alimentary line of com- 
munication, and another of respiration and still others of 
circulation, nervous circuits and the rest; so in a building 
there must be the general line of communication between 
the various apartments, with usually a minor line for service, 
whether the ordinary domestic service of a house, or such 
service as the receiving and cataloguing department of a 
library, or the medical stores supply of a hospital. 



These lines of communication form the basis of rational 
as well as of aesthetic planning, and of rational as well as 
aesthetic external composition. 

The simplest possible plan of a building corresponds to 
that of the simplest possible plan of a living being — a cell 
with a single opening which serves for both exit and entrance, 
for both the users of the building and the service. 

Such was the ancient temple, and such have temples in 
general continued to be, together with other halls of assem- 






"I r 


-1 — 1 

H, r i£ M 





Rational connection between fundamental principles of planning and 
aesthetic principles of external design. 

blage, with a continual tendency to increased proportion 
of the separate function of service, until in the modern 
theater or opera house the problem is to subordinate 
aesthetically the stage portion, which contains the service, 
although it immensely preponderates in bulk. 


Q Q 
D Q 

Q Q Q 

Q D Q 

Q G Q 


Q Q 
Q D 

As soon as we leave the great, single-roomed building 
and come to those of several rooms the line of communica- 
tion between them becomes necessary. 

In its simplest form it is a corridor with the entrance at 
the middle of its length — the central point, from which all 

parts can be reached 
with equal ease, a, 
Fig. 126. 

When the number 
of rooms is so great 
that the corridor be- 
comes inconveniently 
long, the most distant 
parts may be set at 
right angles, as at b, 
thus bringing them 
within a more con- 
venient distance from 
the entrance, and 
forming the well- 
known H plan, or as 
at c, which is substan- 
tially half of the same 
These correspond precisely to the fundamental types of 
external design, the single and double primary masses. 

In both of them the entrance is where it naturally should 
be, both logically and aesthetically, as near as possible to 
the main artery at its central point. If two entrances are 
required, for different uses, as for a main entrance and an 
entrance for service, they may best be placed, for both 
practical and aesthetic reasons, in the masses themselves, 
at the ends of the corridors. 

Fig. 127. 
Further development of previous illustration. 



A further development is possible to meet further require- 
ments, as shown at b, Fig. 127. In this, after entering at the 
central door, as usual, and going a certain distance toward 
either hand, we reach a point whence three corridors diverge, 
bringing all the apartments in connection with them within 
equal distance of the point of divergence. 

This corresponds to the double primary mass with two 
appendages, a; or, if we prefer, to the single primary mass 
with two secondary masses, as at c; or to a single primary 
mass with two appendages and two secondary masses. 

In passing from these to the arrangement of com- 
munications corresponding to triple primary masses, a 
difficulty is encountered which forms a curious parallel to 
the aesthetic hesitation arising at the same point, suggesting 

Fig. 128. 
Modifications of plan that imply a triple external arrangement. 

the possibility that the foundation of the aesthetic sense 
may ultimately be identical with that of the necessary 
physical constitution of the objects of that sense. 

If we lay out such an arrangement as at a, Fig. 128, 
with the view of making one which shall be analogous to that 


at 6, Fig. 126, and shall in addition correspond to a triple 
primary mass, we find that we have lost the logical idea in 
plan entirely. We find that, so far from securing short and 
easy communication, we have readied a most involved and 
roundaboul arrangemenl ; we cannot even reach the central 
poinl of the building without traversing the central corridor, 
and from the entrance to any extreme point, we are obliged 
to 1 >ass over three sides of a rectangle. 

It is only when the central mass is of a different kind, 
either a more important part of the building, a nucleus of 
some sort, whether a "hall of wasted steps" or a many- 
chambered pavilion, that in itself constitutes the center of 
the building, that the triple mass becomes logical (b). 

In such a case the central pavilion demands a different 
treatment from those flanking it, as much practically as 

The alternative is shown at c; it consists in shorten- 
ing all three of the cross corridors so much that they almost 
cease t< 1 1 >e c< >rrid< >rs, and become mere offsets from the main 
c< »rridor. 

In this case the three pavilions admit of similar treat- 
ment externally, with the same strong tendency of the com- 
position to revert to a single primary mass with three 
secondary masses that the plan shows to revert to the 
single corridor with which we began. 

These comprise many other arrangements which need 
not be separately enumerated; thus, an arrangement around 
a central open court is on the exterior suggestive of 
nothing more than the single corridor, or the rear view of 
Fig. 126, c. When any arrangement analogous to those of 
which we have spoken is under consideration, the external 
treatment at once suggests itself, but the convenience of 
having the possible variations of these labeled and pigeon- 


holed in our minds is very great, and much time in fishing 
for a motive and arguing in favor of its practicability may be 
saved if some of the impossibilities are known and recognized 
in advance. 

There are, however, many cases in which the lines of 
communication offer no suggestion as to the proper external 
treatment. Indeed, one of the most serious deficiencies of 
the utilitarian school of design is that only too often the 
most natural and straightforward composition is a perfectly 
plain parallelopipedon, with roof either flat, or of just 
enough pitch to shed the rain. Such was the cella of the 
ancient temple, in its simple form, and without the peristyle 
that was sometimes applied for its adornment. And very 
often in designing a modern building, whether a dwelling 
or a commercial structure, we find that everything naturally 
falls into the same oblong plan, with no need of any projec- 
tion which might give a clew for the exterior. 

When this happens we are forced to apply suitable 
architectural objects, whether secondary masses or details, 
according to the rules that we have been discussing. 

Sometimes the exigencies of the plan require some 
special projection or other peculiarity which in turn will 
lead us directly to a natural and pleasing exterior, but 
frequently we are left with no such aid, and must invent a 
treatment to suit the conditions. 

Take for example a case that is shown in Fig. 129. A 
country house of modest pretensions was comprised in a 
plan 36 feet square. There was no reason for making a 
material projection anywhere. The rooms were just right 
in size and relations. The result was at this point the 
packing box at a. 

It would have been quite practicable and quite proper 
under some circumstances to have treated this as shown at 


with a portico placed at its front. The surroundings, 
however, would not have harmonized with such a scheme. 

It was a rustic, rural 


i |__J I 

place, with wild woods 
about ; the few houses in 
the vicinity roofed with 
pitched, shingled roofs in 
the most unpretentious 
cottage fashion, quite out 
of keeping with prostylar 

As for a plain, straight- 
forward hipped roof on 
our packing box, it 
proved unavailable, quite 
overpowering the house, 
and making the roof 
space of little use for the 
two or three servants' 
rooms that were desired. 
A plain, straightfor- 
ward gable was rejected 
for some of the same rea- 
sons, nor did a mansard 
seem to be quite what 
was needed ; it became apparent after a while that a simple, 
single mass treatment of any kind would not look well. 

The attempt was next made to divide it into two primary 
masses. This could not be done in the ordinary way as at 
g, as the frontage was not sufficient. It became necessary 
to dispense with the link and to join the masses by partial 
fusion, so to speak. It is rarely allowable to place two 
complete gables in contiguity as at h, as this gives the dis- 




Fig. 129. 

Modes "f treating a simple plan by the 
addition of secondary masses. 



agreeable effect of lack of connection commonly called 
"double composition." 

The result was as is shown at d, in which little octagonal 
oriels have been added as two secondary masses. To still 
further soften any stiffness 
of effect a small octagonal 
bay was placed as a secon- 
dary mass at one angle, with 
the roof of a complementary 
pitch, and a pavilion, also 
polygonal, at the opposite 
angle of the veranda, giving 
at last a fairly satisfactory 
result ; and one which as- 
tonished the owner, who 
wondered what was coming 
when he saw the packing 
box in frame. The horizon- 
tal line of the veranda is dis- 
tinctly detrimental, but it 
could not be dispensed with. 

Another example is 
shown in Fig. 130. In this 
case, although the plan of 
the house is a simple paral- 
lelogram, the naturally sym- 
metrical disposition of the 
principal rooms offers a hint 
of a suitable treatment. 

A bay on each side, car- 
ried through both first and 

second stories, seems to promise well. It will be observed 
that the single primary mass of the house is not, as in the 




ious example, treated as a double primary mass. On 
the contrary, it is allowed to remain unmodified, and the 
two bays are planted upon it as secondary masses. 

At the first attempt sonic difficulty was encountered in 
harmonizing the dimensions of the bays with those of the 
front. The width of the rooms at the first laying out (b) 
was not sufficient to permit bays of breadth proportionate 
to the trout, and the comparatively tall and narrow pro- 
jections that resulted were eminently unsatisfactory (a). 
But by modifying the rooms themselves, increasing the 
width and diminishing the depth, space enough was found 
at last f< >r the necessary breadth of the bays (d) . This proc- 
ess not only permitted broader bays, but at the same time 
increased the length of the whole front, thereby still further 
assimilating its dimensions to those of the bays. Finally, 
by lowering each story a trifle, and bringing the eaves as 
far down as practicable, a satisfactory result was obtained (c). 
A circular plan was given to the bays as a contrasting 
treatment ; a square or octagonal plan might have been used 
with good, but different, results. 

Another motive which is very often available is the 
-ingle secondary mass, whether suggested by the plan, or 
attached as an ornament, with some reasonable excuse if 

Take the case, shown in Fig. 131, in which the bald 
conditions of the plan are shown at a, the projection in 
front almost forcing itself upon us, no matter how we 
turn and manage the arrangement in an effort to get 
rid of it, for our preconception intended quite different 

It seems better, finally, to accept it as a necessary datum, 
and to make it the motive of the composition, which works 
out at last as at c. The projection we have carried up as 



a secondary mass, in this case an octagonal hay, contrasting 
with the main roof of the house. The difficulty here en- 
countered lies in the fact that this projection is Hush with 
the end wall of the house, a position in which it is always 
hard to preserve its character as a secondary mass. This 
is partly overcome by giving a slight batter to the walls of 
the bay, thus secur- 


ing a break of four 
inches in the second 
story. The veranda 
at the south end we 
also make a similar 
octagon, and we place 
small octagonal 
hoods on each gable, 
which serve an addi- 
tional purpose in pro- 
tecting loophole win- 
dows that we wish to 
introduce in order to 
obtain a current of 
air through the va- 
cant roof space above 

the attic. The completed composition constitutes a single 
primary mass, with a secondary mass, the octagon bay, 
upon it, and with two asymmetrical appendages, the ex- 
tension and the veranda. 

Such a secondary mass may be set asymmetrically, as in 
this case, or as the central feature : in either case, it is es- 
pecially important that it be kept duly subordinate, as 
otherwise it is apt to assume the appearance of being itself 
the primary mass, reducing the parts of the building on each 
side to mere appendages. 



Fig. 131. 
Treatment with a single secondary mass. 


Thus in Fig. [32, </, the pedimented projection is un- 
satisfactory: ii is either too large or not large enough. It 
for a secondary mass; the eye hesitates whether 
to regard h as primary and the 
parts ( in each side as appen- 
dages, or to regard the pedi- 
mented part as a secondary 
mass, sulx irdinate to 1 he pri- 
mary mass of llio building. In 
such a c.isc much depends up- 
on the projection of the cen- 
tral part; if the projection be 
considerable the subordination 
in height should be greater, is, the more it projects 
in plan the lower should the 
central object be in relation 
to the main building. What- 
ever the projection, such an 
arrangement as a is unfortu- 
nate. It is better either to 
increase the size of the central 
1 1 1; >. part and insist upon it as the 

Relations of pedimenl to build- primary mass, deliberately re- 
ing are uncertain at a, dominant at ... , a , . 

ordinate ducing the flanking parts to 

appendages as at b; or to re- 
duce the size of the central part, as at c, in order that its 
subordinate character as a secondary mass maybe unmis- 

To the mind of the designer who once becomes ac- 
customed to the classification of parts of buildings and their 
combinations thai we have proposed, the various groupings 
me a coherent series, to be looked over and one or the 


other selected, with as intelligent a purpose and as definite 
knowledge of their capabilities as is displayed by a carpentrr 
in selecting from his chest a ripsaw or a backsaw. 

The number of possible motives is exceedingly limited, 
but the number of their combinations is illimitable, as is the 
case in the works of nature. 

They are substantially the same that we set down in 
classification of buildings in Chapter VIII, but it will be 
well to recount them here in slightly different form. 

There is first the question of primary massing, which 
means the question whether we are to let the building alone 
as a single mass, or to try to divide it into two or three 
masses ; and, if two, whether symmetrical or asymmetrical. 

If we determine to let it alone, the next question is 
whether we shall adorn it with secondary masses, or again 
let it alone and treat it with details only. 

The third step is to determine whether to apply a single 
secondary mass, or a pair, symmetrical or asymmetrical, or 
three; if indeed we have not adopted the alternative course, 
and let it alone for detail treatment only. 

At each step, both of primary and secondary massing, 
appendages may be attached, either one or two, and the 
latter either symmetrical or asymmetrical. 

This is absolutely all that is possible as far as massing 
is concerned, but, with these simple elements, what an 
infinite complexity of combinations may be compassed! 

It may seem a bold statement to assert that any classi- 
fication can contain all of the aesthetic possibilities, that 
feelings can be tagged and labeled and inventoried. 

The only appeal is to the already experienced and skilful 
designer. Let such a one observe whether what he is in 
the habit of doing by instinct is not included in one or the 
other of these motives; let him endeavor to do something 


quite different from anything here outlined, and he will 
find that nothing else can be devised that is not incoherent 
and unintelligible, although perhaps romantic or picturesque, 
the latter qualities often existing where unity in the whole 
is wanting. Thus a collection of buildings may be mani- 
festly a mere collocation of separate structures, yet the 
general effect of the whole may be picturesque and pleasing 
although no one would think of calling it a composition, in 
any but the broadest pictorial sense. On the other hand, a 
composition which is agreeable and yet lacking in unity 
usually presents the appearance of several different build- 
ings standing elose together. 

Such are often the picturesque groups of buildings 
erected at various periods of which many English houses 
are composed; and it is difficult to separate the charm of 
romance, antiquity, and spontaneity in which they abound 
from that of coherence of motive, the true test of excellence 
of composition, in which they are often entirely lacking. 



VERY different from the picturesque but unorganized 
combinations alluded to at the close of the previous 
chapter, is organic but asymmetrical grouping, and by 
asymmetrical, we mean just now the visibly and con- 
spicuously asymmetrical. 

Professor Goodyear has shown, among his other brilliant 
discoveries, that exact symmetry, even in that which pur- 
ports to be symmetrical, is not so pleasing as slight diver- 
gences from precise mathematical equality of the two sides. 
As soon as this is mentioned it recommends itself to the 

Such a design as that at Fig. 130 c is wonderfully improved 
if small differences are introduced. If one of the bays is 
fifteen feet across, make the other fourteen, and, keeping 
the roofs of the same pitch, let the ridge of one work out a 
few inches below that of the other. Make the room in which 
the wider bay occurs a foot or two wider than the cor- 
responding room, thus bringing the entrance out of centre, 
and place the central window above it still a few inches 
more off centre. These divergences will be distinguish- 
able only to a critical professional observer, and will often 
be overlooked even by him, while they will add incredibly 
to the softness of effect of the completed building. 

In the house shown in Fig. 60, one of the oriels is a 
foot wider than the other, the octagonal veranda and 



Fig- 133- 
Primary mass with one appendage. 

window above are both slightly off centre, yet no one notices 
the irregularity, and the building is generally liked. 

But we are speak- 
ing now of an en- 
tirely different sort 
of asymmetry — of 
the conspicuous and 
unavoidable kind, 
in which one part 
does not pretend to 
have an answering 
part at all, or none 
that can for a moment be supposed to be its equal. 

The simplest of these asymmetrical motives, and one 
which occurs in all styles, and in all varieties of buildings, 
from the most rustic to the most polished, is the combina- 
tion of a single mass with a single appendage. 

From Fig. 133 to 136, inclusive, are given several ex- 
amples, as diverse as possible in their uses and dimensions. 
In the first, Fig. 
[33, a stable, the 
part on the dexter 
side, that is, toward 
the left hand of 
the observer, with 
the hipped roof, ris- 
ing to the highest 
ridge, is the primary 
mass; the part with 
the long and much 

Fig- 134. 
Gate Lodge, North Easton, Mass. 

Single primary mass with one appendage. 

lower ridge, running off toward the right, is the appendage. 

[34 is a specimen even more rustic in execution, 

although devi ited to similar purposes. It is a gate lodge by 


J 93 

Richardson, and it is built upon precisely the same motive, 
hipped roofs and all, only here the primary mass is at the 
observer's right, the appendage toward the left. In both 
this and the previous example there are various secondary 
masses which do not affect the main composition. 

A much more pretentious building is shown at Fig. [35, 
a Chamber of Commerce building, in which the same mo- 
tive is unmistak- 
able. Here, how- 
ever, the roof is flat, 
of the mansard 
type, the fact that 
the appendage is an 
appendage being in- 
dicated by a break 
in the roof, as well 
as in the plan, and 
by the difference in 
the details, three 
large dormers be- 
ing used upon the 
primary mass, with 
only a half-sup- 
pressed one on the 
appendage. Al- 
though very differ- 
ent in character, it 
is evident that the general motive of the composition is 
precisely the same as in the two previous examples. No 
attempt is made to treat both sides symmetrically, although 
the building is in a style in which much stress is laid upon 
the importance of symmetry. 

Still another specimen of this motive is shown at Fig. 

Fig. 135- 
Chamber of Commerce, New York. 

Single primary mass with one appendage. 



1 56. Although devoted to perhaps less dignified uses, this 
fax surpasses any of the previous in size, being a building 
of many stories, the abode of many families. The primary 

mass is at the left, the ap- 
pendage at the right. The 
former is distinguished by 
a symmetrical treatment 
throughout, and by a very 
large, two-storied dormer 
window. The appendage 
is set off by a slight break 
in the plan ; it has no 
corresponding part on the 
other side; and, most no- 
ticeable of all, the differ- 
ence in the roofs is marked, 
the line of the ridge of 
the appendage being much 
lower, and a dormer of only 
a single story being placed 
upon it. It might have 
been possible to omit even 
this, or to reduce it to a 
mere bull's-eye, or pair of 
bull's-eyes, but for the real 
or fancied convenience of 
the tenants, who usually 
expect full -sized windows. 
The last specimen of 
this motive that we shall 
give is shown in Fig. 137. 

Fig. 136. 

Hotel Essex, New York. 

Still another single primary mass, 
with one appendage. 

It is especially intended to illustrate the adaptability of an 
asymmetrical motive to a large and involved composition. 



At the right is the primary mass, with a long, straight 
ridge, elaborated with a great tower, broad but low, to 
harmonize it with the general horizontal treatment, which 
constitutes a secondary mass. This tower is indeed so large 1 
that it transcends the limit of size allowable for a secondary 
mass, and almost asserts itself as the primary mass. That 

Fig- 137- 
House at Kansas City, Mo. 

A large and elaborate example of asymmetrical composition. 

it is really secondary may be tested by tracing the plate and 
omitting the tower, carrying the horizontal lines through 
and substituting details, dormers, perhaps, and a chimney, 
where the tower was. It will be found that the building 
retains its identity perfectly, demonstrating that the tower 
is really secondary in importance to the general outline. 

At the left of the spectator is the appendage, almost 
as long, but not nearly so high, as the primary mass. The 


line at the top of the first story runs through, but the cornice 
drops considerably below the main cornice, and the ridge 
still more below the main ridge. 

The really important point is that symmetry, although 
pleasing, is not necessary to a high type of design. 

Symmetry indeed may be compared to rhyme in poetry, 
one part made to answer clearly to another, honey-sweet 
indeed, but needing the hand of a master to avoid cloying 
our taste with its charm. 

Or it may be compared to the quatrain stanza construc- 
tion in music, in which the grandeur of a "Dies Iras" may 
indeed be attained, but which must be transcended if a 
"Gotterdammerung" is to be accomplished. 

Next in ( >rder to the motive that we have been discussing, 
is that of a single primary mass with two asymmetrical ap- 
pendages, of which an example is shown in Fig. 131, the most 
sent >us defect being the disproportion between the appendage 
on the left and the primary mass. A longer and lower 
appendage is required, but was unattainable. 

Both single and asymmetrical double appendages may 
be used with double or triple masses, of which one ex- 
ample is given at Fig. 138. This is an extremely complex 
and interesting group of two primary masses, the two 
main gables connected by a link, with the ridge line kept 
carefully a few inches below those of the gables. Upon 
this is placed asymmetrically a secondary mass, the octagon 
bay, with roof line harmonizing in elevation and contrasting 
in plan. That this is secondary in relation to the whole 
composition and not merely to the contiguous gable is 
evident, first from its size, which equals or even surpasses 
the gable in general effect of height, and also of pro- 
jection ; secondly from the care with which it has been 
separated from the gable by the vertical line of the chimney, 



instead of being joined to it by making the horizontal line 

The farther gable, on the other hand, is itself a complex 
group of two unequal masses, a smaller gable being attached 
and bound to it by partly lapping over upon it, and by in- 
sistence upon the horizontal line at the second-story window 

To this unusually involved group are added two ap- 
pendages, the farther one much smaller than the nearer; 
but the roof even of the nearer has the ridge kept down 

Fig. 138. 
Two primary masses, with two asymmetrical appendages. 

considerably below the ridge of the link. Notice, too, the 
way in which the dormers are placed upon this nearer ap- 
pendage, a large and a small, assimilating themselves to the 
unequal gables of which the farther primary mass is com- 
posed. All this was done quite unconsciously of any 
purpose, and solely by the instinct of the designer. If we 
had asked him, " Why do you make this ridge lower than 
that one?" he could have made no reply, but that he 
thought that it looked better so. 

But when we find by examination of many examples 
that this is but one of many similar cases, in buildings of 


widely varying types, we arc justified in making a general 
statement, and inventing a suitable nomenclature of 
classificatii >n. 

This constitutes the reduction of architecture to a 
science, for all science is but the classification of facts and 
generalization of their relations. 

Besides the combinations with a single appendage and 
with two appendages, a most useful asymmetrical com- 
position is that of two unequal primary masses. 

Fig- 139- 

House at Detroit, Mich. 

Two unequal primary masses. 

It might be justly said, indeed, that this is the funda- 
mental type of such compositions, as it is the only group 
of unequal primary masses which has been generally used. 

Wo have .already given illustrations of compositions of 
two unequal masses, both primary and secondary. We now 
add a few in addition to show the flexibility of the type, and 
its easy adaptability to widely varying conditions. 



Two precisely parallel compositions are shown in Figs. 
139 and 140, although the difference in the forms of which 
they are composed is complete. The contrast between the 
hipped roofs of 
the one and the 
broad, flat gables 
of the other is 
not more strik- 
ing than the ab- 
solute analogy 
of their arrange- 

Even in the 
minor parts a 
curious likeness 

Fig. 140. 

Two unequal primary masses, a close parallel to 
Fig. 139. 

prevails. Note the great central chimney in the middle of 
the front of the larger mass in each ; and in each the bay, 
projecting forward as a single secondary mass, at an exactly 
corresponding point in plan. How easily could one be 

changed into the 
other, without dis- 
turbing in the least 
the interior do- 
mestic affairs. The 
styles are different: 
the motive is iden- 

Another example 
is given at Fig. 141, 
to show the avail- 
ability of the motive 
for less rural structures. In this case it well interprets the 
uses of the building, as the smaller mass marks the less 

Fig. 141. 

Two unequal primary masses. Note the dis- 
proportion between masses and link. 



or as here, 
may be en- 

important entrance, which leads only to the public hall in 
the second story, indicated by the three large windows; 
while the Larger corresponds to the main entrance to all 
the rest of the building. 

One or two other points are illustrated in this example. 
The first is that it is not necessary to have either the entrance 
or a central feature placed upon the link; the entrance may 

be as well in either 
trances in both. 

The second re- 
lates to proportion. 
The second story of 
the link, in which 
occur the three win- 
dows of the audi- 
torium above men- 
tioned, should have 
been farther back, 
on a line with the 
wall of the upper 
story. The space, 
however, could not 
be spared. The result is seen in the too massive, square- 
shouldered appearance of the whole. It is caused by the 
that the rectangle of the link between the lines of the 
towers and up to the point where the wall stands back, is 
quite dissimilar to the rectangles of the two towers. 

Another defect in proportion, of less magnitude, is in 
the relation of the front faces of the octagons above the line 
where they begin. Although the cornice line of the smaller 
tower has been lowered a little, it has not been lowered 

lliw pi i 


Fig. 142. 
Two unequal secondary masses. 


enough, and the effect of a height greater than that of the 
larger tower still remains and grates upon the sense of the 

The last of this type that we shall give is at Fig. 142, a 
sketch for a building that was ruined in the execution. 
The octagon tower on the corner comes naturally enough in 
connection with the plan as the proper place for the entrance 
to the banking room in the first story. The upper stories 
are laid out as offices, and the entrance to these at the side, 
three-quarters of the way toward the rear, is marked as 
naturally by a reflection of the larger tower, built on purely 
for adornment, although eight inches was the greatest 
projection that could be obtained. 

Both towers are so much subordinated to the main roof 
of the building, that they can rank only as secondary masses, 
the first instance of the kind that we have discussed. 

We might go on and show examples of variously com- 
plex arrangements of secondary asymmetrical masses, with 
their asymmetrical details and appendages, but enough 
has been said to suggest to the student the point of view 
from which he may analyze for himself the most invoked 



THE practical work of designing a building consists in 
adapting the plan to a suitable external motive; or 
in adapting the external motive to the plan, whichever 
ment may be preferred. In point of fact both opera- 
tions proceed simultaneously and almost instinctively; in 
this work a clear conception of the connection of the various 
possible types of exterior compositions is of much assist- 

Each type may be regarded as derived from a preceding 
type and all may be traced back to the primitive single 
mass type, from which the more complex forms are con- 
ceived to be built up by a process of accretion, the addi- 
tional members being added consecutively, one by one. 

If we begin with the simplest kind of building possible, a 
parallelopipedon with door, windows, and cornice — a single 
primary mass — Fig. 143, a, the first modification that is 
possible with parts of the same order is the extension of a 
single appendage, as at b. 

This gives us an asymmetrical group, and one which, 
as we have seen, is used and approved by all schools of 

When the exigencies of the plan require it, this ap- 
pendage is separated from the primary mass, remaining 
connected with it only by a link, as at c. 

The beginning of this process may be seen in Fig. 115, 
in which the link has as yet hardly become a link, the 



appendage has barely become detached ; indeed this building 
has been examined heretofore without any suggestion of 
this incipient separation. 

At c the separation is complete, and a composition of 

a a o 



a a 

a a 

a a 




a o 

Q I S 

q a a 


D Q 

a a 

D Q Q 

a a 



q a 

a a 

a q m 

q n 
Q Q 




Q Q 


uj) a 1 a 



a qq 





Q ll u a QDfl 
Q Q u I u Q I u f ) u 

D U u 

uii 1 1 Q ■/ 


a q 

ft- 4 


Q Q 

LI u D 



Fig. 143- 
Development of one type of composition from another. 

two unequal primary masses, connected by a link, is the 
result. Any distinguishing secondary mass, such as a turret 


or belvidere, is naturally placed upon the largest of the two 
masses, or on both of them, bu1 a distinguishing detail, such 
as a doorway, may be placed cither upon the masses or the 

If the smaller of the two masses be increased in size it 
ultimately becomes equal to the larger, and a group of two 
equal masses is formed (d). In this stage the doorway, or 
other tertiary part, may still be upon either or both of the 
masses, or upon the link, but a tower, or secondary mass of 
any kind, if placed upon one mass, must be repeated on the 
other also. Usually, however, such secondary mass is 
better placed upon the link, as at e, especially when the link 
is increased in height to that of the masses, which is not 
shown in the diagram. As the link grows in importance, it 
may itself become the central mass of a group of three, 
the flanking masses being now separated from it, remaining 
attached only by links, as at /. 

Going back to the beginning at a, we may make the 
first addition two appendages as at g, instead of one, and 
these may be either symmetrical as to their size, or asym- 
metrical as we may prefer, of which only the former type is 

At h, which is the next stage of development, the ap- 
pendages have been moved away from the primary mass 
and have assumed an individuality of their own, attached 
still by links, and still remaining inferior in size to the central 
primary mass. Here again only the symmetrical type is 
shown, although the flanking masses may be asymmetrical 
in size. Examples of this, however, are rare. Whatever 
contrast in size is admitted, substantial similarity in shape 
between them must be retained. 

In this condition the composition is entirely analogous 
to the corresponding two-part group at c. Both are com- 



plete and satisfactory, and capable of great variation in the 
size of the added masses in relation to the original mass. 

As long as they remain smaller no especial distinction 
of form is felt to be necessary ; the distinction in size seems 
to be sufficient to concen- 
trate the attention on the 
largest and to give the re- 
quired unity of impression. 
It is true that if any detail 
is needed, as a doorway, it 
is most naturally placed on 
the largest mass, but be- 
yond this the eye requires 
a general similarity of form 
in all of the masses. When 
equality of size is reached 
as at i, a sense of dissatis- 
faction occurs: the three 
masses stand apart and 
need to be in some way 
drawn together. We try 
the addition of a secondary 
mass, as at /, just as we did 
in the case of the double 
group, with a result only 
partly satisfactory, as a 

curious effect at first follows the experiment. The addition 
of the turret, instead of giving apparent additional size and 
dignity to the central mass, actually causes it to appear 
relatively smaller, and every additional detail seems to 
diminish still more its size. We are compelled finally to 
restore the preponderance in size to the central mass to 
enable it to support the added enrichment, reaching a final 

Fig. 144- 
Porte des Cordeliers, Loches. 

The two turrets are unmistakably 
secondary masses. 


stage at / identical with that obtained from the double 

gr< >up. 

The combinations shown in these examples comprise 
all the fundamental arrangements of masses that are pos- 
sible. To all of them, except b and g, appendages maybe 
added, which may again be separated by links, giving rise to 
another series of compound compositions, entirely analogous 
to the original series, but of which each individual part is a 
composition in itself of one of the types here shown. 

An equally interesting and profitable series of changes 
and ci »nned i< >ns maybe shown in the conversion of secondary 

masses into pri- 
mary, and the re- 

At Figs. 144, 
145, 146 are shown 
three buildings 
illustrating the 
growth of two sec- 
ondary into two 
primary masses. 
In the first there 
are two turrets, 
springing from 
corbelling nearer 
the top than the 
base of the rect- 
angular building on which they are placed. They are alto- 
gether secondary in character, their bulk so inferior that 
they might be removed entirely without any loss of identity 
to the building. 

A1 Fig. 145 a pair of the same sort of round tourelles is 
increased considerably in relative size, yet still unmis- 

Fig. 145. 
Chateau he Chenonceaux. 

Although increased in size the two turrets are 

still secondary masses. 



takably secondary, and quite subordinate to the primary 

Finally, at Fig. 146 the round towers have reached their 
full size and have in turn become primary masses, the build- 
ing proper having dwindled 
to a mere link between 

This change of second- 
ary into primary masses is 
of continual service in the 
designing of modern domes- 
tic work ; and the conditions 
of plan that require one or 
the other very soon become 
a matter of instinctive feel- 

It is very difficult, for 
instance, when the plan re- 
quires such a projection as 
that shown at a in Fig. 131, 
flush with the end of the 
building, to make anything 
of it as a composition, but 
a single primary mass with 
an appendage. If this is 
done, the difficulty then encountered is that the appendage 
is too large to be properly subordinated to the mass, the 
latter being fixed by the limited width of the projection. 

Whereas if a break can be arranged in the plan, as at 
b, between the projection and the body of the building, 
there is no difficulty at all in treating the projection as a 
single secondary mass as is done in the final result at c. 

When the secondary mass is a central one, a gradual in- 

Fig. 146. 

Chateau de Chaumont. 

The turrets are here increased in 
size until they have become primary 



crease in size (ends to change the primary mass of the 

building into two appendages. 

This process is shown in Fig. 147. At a the colonnade 

is altogether subordinate, unmistakably a secondary mass 

upon the primary 
mass of the building. 
At c the colon- 
nade has grown in- 
to the primary mass 
itself, the building 
being subordinated 
as two appendages. 

At b is an inter- 
mediate stage, in 
which the subordi- 
nation of one part or 
the other is not suffi- 
ciently well marked 
for a really satisfac- 
tory result, it being 
hard to say which 
is meant to be pri- 
mary, the building 
or the colonnade. 

Although this in- 
termediate form is 

Fig. 147- 

Varying relative importance of colonnade from 
secondary at a to primary at c. 

frequently used, it will be found that it always makes a 
more spirited composition if it is changed so as to resemble 
</ or c, in which the character of the central mass as either 
primary or secondary is unmistakable. 

[f the central part of the facade is advanced, while the 
cornice line is kept uniform, as at b, it is apt to detract from 

importance of the whole building without sufficiently 


atoning for it by any added importance of its own. It is 
far safer either to run the cornice line through on a straight 
front, letting the projection top out below it as at a, or to 
boldly carry up the projection, making that the principal 
mass of the composition, as at c. 

With a little practice, it will be found that all of these 
things will take care of themselves in the study of the pos- 
sible variations of a composition. Indeed, as soon as the 
sentiment of the laws of composition is acquired, its execu- 
tion becomes spontaneous and unconscious, as is the case 
with all arts after once the skill has been gained : the dancer 
never thinks of the position of his feet, nor the musician 
of his fingers. 

To show how completely one thing follows another as an 
almost necessary implication, observe the stages of change 
in the same composition that are shown in Fig. 148. At b 
is the same house as at Fig. 60, only drawn sketchily to 
compare on equal terms with its derived forms a and c. 
This design, it may be mentioned, is of peculiar interest, as 
it is the first design that was made through the conscious 
application of the various laws of composition that we have 
formulated, although many had been made with a dim, 
gradually awakening apprehension of the fact that there 
were laws. 

As it stands, it forms a composition that is actively 
pleasing to almost everybody, consisting of a horizontal 
single primary mass, with two secondary masses in the 
upper story and one in the lower. The intention was to 
give it as marked a horizontal character as possible. The 
most serious deficiency, felt to be so at the time, is the 
shortness of the main ridge. This might have been much 
improved by making the pitch of the end slopes steeper 
than that of the front slope of the roof; but country 



carpenters have an invincible prejudice against such trifling 
with the regular mitre line and the unevenness of the 

shingle courses that 

Now, if it should 
be desired to keep 
the same motive in a 
building of more in- 
dividual, vertical di- 
mensions, a general 
extension vertically 
must take place. The 
pitch of the main 
roof must be raised ; 
that of the oriel and 
porch roofs raised to 
match the main roof ; 
the corbelling under 
the oriels prolonged 
to match the roofs, 
and the central de- 
tail that connects the 
oriels modified to suit 
the more contracted 
space between them. 
The process ends 
in a perfectly har- 
monious building of 
tower -like appear- 
ance, though hardly 
available for modern 
use as it stands, ex- 

Spontaneous modifications that accompany , 

fluctuating dimensions of a building. cept perhaps as a 

vi__=-=- - ■ «-.. .'«. 

O hhh rfcj I 

iiifi jipf • 



pavilion of a larger composition, of which it might be a 

On the other hand, if the conditions, such as additional 
land that may be used, larger rooms needed, lower ceilings 
and a more rustic character wanted; if such conditions re- 
quire a more marked horizontal character than that of b, 
the result is exactly the opposite. The main roof pitch is 
lowered as much as we may venture, the oriel roofs are 
flattened as much as possible, and the porch itself is aban- 
doned in order that the marked individuality of a con- 
spicuous single secondary mass may not conflict with the 
continuity of the whole composition. The window details 
that connect the oriels are lengthened out horizontally 
and are made of even number toward the same end, ter- 
minating in a structure well suited for a cottage, and just 
as harmonious in itself as either of the previous modifications. 

Both the vertical and the horizontal lengthening-out 
process is done without a thought of exact measurements, 
of precise equalities of ratios, or of any of the other labori- 
ously extracted qualities that we have analyzed. 

All goes on cheerfully and hopefully, with the aid of a 
soft lead pencil and plenty of tracing paper, only the eye and 
the feelings guiding us, except possibly in some special 
contingency when the solution is not spontaneous, or in 
the case of a larger and more complex building, when a few- 
rules to illuminate the artist's instinct become as useful as 
are now and then the rules of perspective to a painter. 



EVEN for those who arc not designers and never expect 
to be, but who wish to know how to criticise a design 

intelligently, the generalizations that we have made in the 

preceding chapters will be of infinite service. 

Generalization means comparison, and comparison means 

observation; and he who has learned to observe and 

compare is in 
a position to 
verify the gen- 
eral statements 
that have been 
made, and to 
criticise ration- 
ally and not 
arbitrarily. The 
power to judge 
comes with the 
power to ob- 
serve and class- 
ify, so that the 
readers of these 
pages who have 
no intention of 
ming designers may at least become capable critics, who 

arc quite as necessary for the prosperity and progress of the 

arts of design as are the designers themselves. 

Fig. 149. 
Rathiiaus, Vien t na. 

The serious defect is the antagonism between the 
il parts and the general strongly marked hori- 
zontal disposition of the building. 



After what has been said, it will be easy for any one who 
has fully digested it to recognize parallel motives in the 
most diverse styles, and to note at the same time the minor 
differences. Soon will come the power t< 1 weigh these differ- 
ences, assign their 
causes, and esti- 
mate their value ; 
so that the bare 
statement " I think 
this is prettier than 
that," which is the 
sum of artistic criti- 
cism to-day, may 
give place to " This 
is prettier than 
that, because of 
such and such char- 

Take, for exam- 
ple, the two build- 
ings shown in Figs. 
149 and 150. 

These differ widely in dimensions, Fig. 149 being a build- 
ing of many stories and heterogeneous uses, perhaps five 
hundred feet in length, with a central spire of three-fifths 
that in height; the other (Fig. 150) a mere gateway, perhaps 
seventy-five or eighty feet in both width and height ; they 
are separated by about three centuries in time and a thousand 
miles in space; yet the designers of these have hit upon an 
arrangement in each case, which, however different in detail, 
is virtually identical in general conception. 

In each case there is a central mass, with an appendage 
on each side. Upon each central mass there is a group of 

Fig. 150. 
Gateway of King's College, Cambridge. 
In composition closely parallel with Fig. C49. 


five secondary masses, spires in one, pinnacles in the other. 
At the termination of each appendage is another secondary 
mass, a pavilion in one ease, another pinnacle in the other. 
The precise correspondence is remarkable. 

As for the minor differences, and they are many, it is 
not our purpose to dwell upon them now, it is the extraor- 
dinary likeness that we wish to remark, — the larger central 

ndary mass, flanked by a pair of spires or spirelets on 
each side, and other secondary masses bounding each 
composition and attached to the main group by the sub- 
ordinate parts that we have called appendages. 

The faults of both are those incident to an impure style. 
The earlier gateway is in the Gothic of the sixteenth century, 
when horizontal details were beginning to displace the 
pointed; the city hall is a nineteenth-century combination 
of horizontal masses with vertical Gothic spires. Purity of 
style is not a thing to be revered or slavishly observed, 

cially where a transition style, the purity of which lies 

in its impurity, is in question. Purity, properly used, 

means that harmony which is secured by the carefully worked 

out similarity of all parts, both masses and details, until 

1 ype dominates the whole. 

Of such a style as that of the period of Francois I, which 
is a mere picturesque grafting of classic details upon a 
Gothic body, it is vain to demand purity as a necessary 
characteristic, while that of the time of Louis XV, however 
objectionable on the grounds of extravagance and lack of 
delicacy, may properly be called pure, as all the parts share 
the same character. 

So in both the above examples the chief criticism will 
be the lack of similarity in the horizontal and in the pointed 
or vertical parts. The five tall, narrow, secondary masses 
directly contradict the general horizontality of roofs and 



mouldings, while there is no army of little pinnacled buttresses 
as in the true Gothic to justify and harmonize the larger 

Another serious defect is the use of five secondary 
masses. This is partly atoned for in Fig. 145 by the differ- 
ence in form of the central one; but in Fig. 144 there is 
not even this to excuse it. The result in both cases is a 
loss of the unity that would be felt if only three were used. 

The next case, for criticism only, not for comparison, is 
shown in Fig. 151, and it is an admirable example of almost 
all possible blun- 
ders in composi- 
tion. The build- 
ing from one 
standpoint is of 
horizontal char- 
acter, complete- 
ly girdled by two 
heavy cornices 
and partly by a 
third. From an- 
other point of 
view the compo- 
sition appears to 
be a vertical one, as it is cut vertically into various parts, 
marked strongly enough to constitute a primarily vertical 
composition, if they were not contradicted by the strong 
horizontal cornices. The fundamental error is here touched, 
in that there is no clear conception of what character the 
whole design is intended to have. 

If it were meant for a horizontal composition the very 
marked vertical break in plan at the corner should not have 

Fig- 15 !• 

Ring Theatre, Vienna. 

All the parts are contradictory 


been so marked; or rather should not have occurred at all, 
nor should the twin tower-like treatment of the front have 
been attempted; if vertical, the vertical breaks should have 
been more strongly marked; the lines of the towers carried 
down to the ground, the loggia which forms the link set 
back, and the projecting part, formed by the break at the 
corner, cut down in height, to reduce it to an appendage, 
and the horizontal mouldings much diminished in projection. 

A double course is always open to the designer in every 
of every composition, and a criticism, to be of value, 
must point out, or at least imply, the alternative. The 
author lias been asked to give an example of a perfect com- 
position : this, however, is impossible; for every composition 
has an alternative, and when a difficulty is reached in the 
working out that seems insoluble, the chances are that we 
are trying in the wrong direction, and that an experiment 
should be made in just the opposite way. 

Assuming that a vertical composition is to be made, the 
first fault is, and it is a serious fault and a frequent one, 
that the central portion, with the great arched loggia, that 
connects the two towers, does not properly connect them at 
all, but is pushed forward as if it were the principal part of 
the composition. It might indeed be made so, but in that 
the tops of the towers must be cut off and various other 
things done to make the central part duly predominant. 

Taking it as a connecting part between two masses, it 
is absolutely essential that it should not advance by even an 
inch in front of the masses which it is to connect. If it 
does, again we have that fatal and hopeless condition of 
not knowing what we want to do, and of course not doing it. 
First, we have sinned by not knowing whether we wanted a 
horizontal or a vertical building; and now we sin by not 
knowing whether it is to be of one or two primary masses. 



Again, we must realize that, to exist at all harmoniously, 
the part at the left may exist only as an appendage, a hanger- 
on, subordinated in height as well as in plan to our main 

It is impossible to pile it up to the same heighl as the 
tower; we must cut it down at least to the level of the main 
cornice; it would be better if it could be cut down still 
further, to the height of the pretty little open colonnaded 
corner porch, which we must try to make the really effective 
appendage, if we can suppress the aspiration of the one in 
the background by even an attic story. And if this cannot 
be done, if the space it incloses is really indispensable, the 
motive of the whole composition must be changed, the verti- 
cal type relinquished and an entirely different one tried. 

As for the large central gable half seen above, we hardly 
know what to do with it. It may not be entirely removed, 
as that part of the working arrangements inside, we know, 
cannot be dispensed with. We must be content if we can 
work it a foot or two farther back in plan, or hip the roof 
back to do away with the impertinent pediment, and raise 
the flanking towers above it by even a little ; get it out of 
sight in some way, as best we can. 

Finally, comes the sin against proportion in the narrow- 
ness of the towers, which has no likeness in the broadness 
of the central motive. 

It might be possible to establish a very passable relation 
of similarity between the dimensions of these towers and of 
the whole first story front below the stringcourse ; but such 
a suspected harmony is partly masked by the superfluous 

Embodying all of these suggestions in a concrete sketch 
we see the result in Fig. 152. 

In this the link stands back a few inches as it must, 


Fig. 152. 

Modification of Fig. 151 to 
a definite motive of two 
primary masses with an appen- 

with an immediate improvement in effect. The towers are 
: a little and the portico removed, bringing towers and 

first story into some approach 
to proportion. The towers are 
lined across by a moulding be- 
tween the first-story string- 
course and the main cornice, 
dividing the height into three 
parts, which bear some propor- 
tionate relation to the part of 
the link above the first-story 

The pediment is suppressed 
out of sight and the appendage 
cut down one story, trusting that what is left of it will not 
be much noticed, and the pretty little arcaded appendage 
is treated in a way that suggests 
reat loggia, giving the added 
charm of similarity. 

If this solution proves unavail- 
able, on account of the lost space 
in the attic of the appendage, or 
f( ir 1 itlier reasons, we must reject 
the whole scheme for a vertical 
composition, and work up the al- 
ternative horizontal arrangement. 
The tower idea must be elimi- 
nated and the roof of the attic 
carried through. This with the 
other necessary modifications is 
shown in Fig. 153. This disposes 
effectually of the troublesome gable; what little may pro- 
ject above the flat roof is too low and too far back to be 

Fig- i53- 
Modification of Fig. 146 
in an opposite direction, to 
secure a single, continuous, 
primary mass, with a single 
secondary mass, and an appen- 


visible from the street, and it is a false system of composi- 
tion that demands the exhibition on the exterior of some 
symptom' of every internal arrangement. Where the internal 
arrangement suggests the neatest and most straightforward 
motive for the exterior, it is naturally the best motive that 
can be adopted ; but where the internal arrangement is so 
involved that the external expression is difficult, some other 
motive may be better. 

In this case, for instance, the gable appears to be the 
front wall of the auditorium of a theatre ; the wall of the 
facade evidently incloses the miscellaneous collection of 
lobbies, corridors, foyers, ticket offices, and such, which are 
indispensable in a modern theatre. If we had the plan 
before us, and were at liberty to remake that also, it might 
be well to attempt to reduce the foyer and other appurte- 
nances to a smaller space, to fill out the break in plan at the 
corner, and to make the auditorium gable the predominating 
motive of the composition. 

But taking the data of the plan as they are forced upon 
us, it is our task now to do the best that we can with them, 
without radical change. 

The next end for which we must strive, is the sub- 
ordination of the appendage. It is impossible aesthetically 
to carry its cornice through on a level with the attic cornice 
of the main building. We must reduce the height of the 
former and increase that of the latter by every means in our 

Accordingly, we keep down the roof of the appendage, 
drop the window heads, remove the parapet, lower the story 
height if necessary. 

On the main mass we do just the opposite, to give all the 
additional height possible. We put on a balustrade, as 
high as we dare, and if necessary raise the whole cornice. 


■ ■ next step is to treat the attic with a continuous 
treatment of pilasters, around which the attic cornice 
breaks. Now is the time and place to put our statues — not 
pinnacles, such as are used in the original design, — which 
learly out of keeping. Along the whole front we place 
them and al< >ng as much of the return above the appendage 
as can be seen, to give the appearance at least of a complete 
row along the whole side. One is placed at each pilaster, 
carrying up the vertical lines of it, all still in due harmony 
with the horizontal feeling which the even horizontal rows 
of b<»th statues and pilasters produce. 

Almost the only semblance of a pinnacle that can be 
used with the horizontal styles is the human figure, and this 
because it is not so very much of a pinnacle after all, but 
more suggestive of the shape of a vase, small at the foot, 
at the shoulder. Even statues are best used in long 
r< iws, where the continuity of the whole is the prevailing 
; . It will be observed, too, that it is difficult to place 
a n >w i if statues immediately upon a heavy, unbroken, main 
cornice. Usually, as in this case, an attic intervenes, cut 
up by ] (ilasters, with its own lighter cornice breaking around, 
and forming an intermediate, half continuous, half vertical, 
podium upon which the row of statues may stand. 

Below the main cornice the building may stay nearly 
unchanged. The central pavilion with the open arched 
loggia falls into due subordination as a single secondary 
mass, and into proper proportion with the whole front. 

In the first story we may allow the portico to remain 
if it must be, as it is not fatal to the composition in this case, 
although not an improvement. We insist, however, upon 
ing the rusticated treatment of the first story through 
the portico also, in order not to interrupt the continuity 
of the motive. 



It will be as well in this case, as in the former, to make 
the open porch in the angle an arched motive in the second 
story, similar to that of the great loggia, and a like treat- 
ment for the windows of the attic will improve the ap- 
pearance of the whole. 

If it were possible to dispense with the attic entirely, a 
third, and fairly satisfactory solution might be reached 1 >v 
removing everything above the main cornice. Cover that 
portion and what remains is a coherent composition. The 
central part then becomes a secondary mass ; and the small 
loggia at the left declares itself as the effective appendage, 
being large enough to throw the part above it into the 
background. In- 
deed the building 
as it stands looks 
as if the attic 
story had been 
added afterwards. 

The next ex- 
ample, at Fig. 154, 
so far from being 
seriously faulty, 
is a fairly cor- 
rect and credita- 
ble performance, 
open only to 
minor comments. 

It is intro- 
duced to show how much better results may be obtained 
from this frequently used arrangement if it is swerved a 
little, one way or the other, in order to give more coherence 
to the parts, that is to say, unity to the composition, for 
which we are always striving. 

Fig- 154. 
Palace of Ludwig Victor, Vienna. 

The conspicuous defect is the lack of subordi- 
nation of the central pavilion. 


As it stands it is composed of a plain, solid, square mass 
behind, in front of which is placed a flat frontispiece, much 
richer than the main building, and extending as high as the 
main building itself. Now the most untutored eye will see 
that this frontispiece causes a serious interruption in the 
line of the main cornice. It is not clear whether the pieces 
of it that appear on each side indicate the boundary of the 
main house, and the enriched frontispiece is merely applied 
to it; or whether the frontispiece itself is the main mass, 
and the parts on each side appendages of it. 

Neither building nor frontispiece is subordinate to the 


Almost all of the errors in composition that are made 
arise fr< >m this very cause, that the designer is not clear as to 
what is to be done and in consequence adopts a doubtful 
middle course, trying to do two opposite things at once and 
failing to get the full effect of either. 

The one rule for composition is that the designer shall 
make up his mind to whatever motive the physical require- 
ments of the plan point, and devote himself to the clear 
expression of it, restudying and if necessary modifying 
features of the plan that conflict with it. 

Wry often such a process results in great improvements 
in the plan itself, and both plan and exterior should seem so 
simple and inevitable when completed, that to the un- 
initiated it will be incredible how many weeks of work have 
been lavished upon them. 

En this d< sign, and it is but one example of many instances, 
for nothing appears more obvious than to advance a little 
the central enriched part of a building, there are as usual 
two modifications, either of which is capable of producing 
a more coherent composition. 

In the first, Fig. 155, the frontispiece has been kept 



• iJsl 5a-52h. 

Fig. 155- 

Modification of Fig. 154, re- 
ducing the central portion to a 
definitely secondary mass. 

down, and made a distinctly secondary mass, in proper 
subordination to the primary. The gain in coheren* 
at once seen. It would be very 
desirable also to lower its cor- 
nice as its proportions are not 
those of the primary ; it is some- 
what too tall in comparison. 
This, however, we cannot ven- 
ture to do without some oppor- 
tunity of revising the plan to 

Ij: we prefer to make the 
opposite modification, we shall 
have some such result as Fig. 
156. In this the central frontispiece becomes the primary 
mass, and the cornice is returned for the full depth of the 
building. From the side parts the upper story is cut off, 
reducing them to subordination as appendages, the roofs 
of which are made similar in character to that of the 

primary mass. 

The coherence again is per- 
fectly clear, clearer perhaps than 
in the first modification ; but so 
great a change is involved in the 
plan, in cutting off so much of 
the upper story that in practice 
this second transformation might 
be unavailable. 

A more serious defect from the 
aesthetic standpoint is the dispro- 
portion that exists between the 
primary mass and the appendages. 

The latter are far too narrow to present any approach 

Fig. 156. 

Modification of Fig. 154, in- 
creasing the central portion 
until it becomes the primary 


milarity with the main dimensions. If one or two more- 
windows in breadth could be added to each, a far more 
harmonious result might be attained. It is impossible, 

h< »wever, to do this without 
L< >sing the entire outline of 
the original building. 

The last example that 
we shall consider is shown 
at Fig. 157: a church, at 
first glance of the usual 
type, with twin towers and 
a gable between. A second 
glance shows that a curious 
dislocation has taken place, 
the gable is pushed forward 
very much in front of the 
two towers, while the towers 
have drawn away sidewise, 
to the greatest extent pos- 
sible without losing all con- 
tact with the building. It 
looks as if a force of repul- 
sion animated the parts and 
drove them asunder, a result 
quite destructive to a sense 
of unit v. which demands their approach and close connection. 
This is the same sort of mistake as one of those mentioned 
in connection with Fig. 151, the pushing forward in plan of a 
part whose only proper function is that of a link to unite 
twi 1 masses, and by its nature to be subordinated rather than 
made prominent: there, however, the pushing forward was 
measured by inches; here it is fifteen or twenty feet, and 
proportionately a greater shock to good taste. 

Fig. 157- 

The serious fault is the prominence 
of the nave and narthex, making it 
too conspicuous for a link, but not 
enough so for it to become itself the 
single primary mass. 



It is an error arising from failure to understand that a 
composition may be one of two masses, or one of three 
masses ; but it cannot be both at once. 

If it is to be one of two masses the central part cannot 
be anything but a connection between them, on pain of 
losing all connection and all sense of unity, as here occurs. 

On the other hand if it is to be three masses, the central 
must dominate either in height or bulk, but never in pro- 
jection, because by such projection one or the other of the 
two remaining masses is more or less cut off and its impor- 
tance as one of three masses diminished ; unless, indeed, the 
two flanking masses are to be relegated to the position of 
secondary masses, and then the central mass may stand in 
whatever position in plan we 
may prefer. 

But it will do little good here 
to place the gable back on the 
line of the face of the towers, 
unless we can in some way obtain 
a better connection between it 
and the towers than is afforded 
by the one-story high, horizontal 
line of the balcony of the narthex, 
which then will become that of 
the front itself. 

This connection may be im- 
proved, it is true, by continuing 
the small arcade across the fronts 
of the towers, and putting an- 
other arched doorway in each 
tower, making five instead of three arches; but there would 
still remain, standing upon this coherent, horizontal base 
the three disconnected upward projections of the towers 

Fig. 158. 

Modifications of Fig. 157, to ob- 
tain a more coherent result. 


and the gable, open to precisely the same criticism as 
bef< 'iv. 

[f to overcome this state of affairs we determine to 
construct a horizontal line across the gable, as is done in 
Notre Dame at Paris, as shown at Fig. 158, we at last gain 
unity, as far as the proper connection of the parts is con- 
cerned, but there is a total lack of proportion between the 
broad mass of the central link and the tall, narrow masses 
of the towers. Nor can any remedy be suggested which 
does not require a complete overthrow of the conditions. 

As for the details of the composition, they are as de- 
fective as gleanings from two different styles, patched to- 
gether by a designer of cold and petty sensibilities, might 
xpected to be. The general motive of the detail is the 
round arch, and the circular plan of the pinnacles is rather 
a happy treatment in connection with them. To complete 
it, however, it should have been carried into the spires 
themselves, the present octagonal plan being without the 
necessary similarity. 



Academy of Fine Arts, Vienna . 128 

Albani Villa, arcade of 42 

All Souls Church 40 

Analogy between horizontal 

and vertical division 127 

Analogy in composition of de- 
tails and primary masses ... 103 

Analysis of buildings 57 

Angle turret, proper and im- 
proper treatment 32 

Antwerp, Palais de Justice. ... 77 
Appendage, composition with 

only one 192 

Appendages 60 

Appendages, compositions with 

two asymmetrical 196 

Appendages, proportions of . . . 144 
Arches and domes, similar 

forms 40 

Arches, harmony with circular 

plan 41 

Arches, inharmonious effect of 

dissimilar 43 

Architecture, relation to con- 
struction 9 

Architecture, scope of 8 

Art, pure and representative. . 12 

Art, unity in 19 

Arts, classification of 13 

Arundel Castle, gateway 70 

Asymmetrical composition. .. . 191 
Asymmetrical compositions 

with one appendage 192 

Asymmetrical composition with 

two appendages 196 

Asymmetrical composition two 

unequal primary masses. . . . 198 


Asymmetrical composition, two 
unequal secondary masses. . 199 

Asymmetrical double primary 
masses 70 

Asymmetrically ] 'laced single 
secondary masses 88 

Azay-le-Rideau, Chateau de . . 90 

Banqueting Hall, Whitehall ... 120 

Bartolini Palace 154 

Bartolini Palace, tabulated 

proportions 153 

Berlin, Comedy Theatre 66 

Berlin, Synagogue 149 

Borghese, Villa 145 

Bourse, Lyons 76 

Bourse, Marseilles 121 

Bridge composed of dissimilar 

arches 43 

Buckingham Palace, London . . 77 
Buildings classified according 

to primary massing 85 

Campanile, Lucca 133 

Campanile, St. Mark's 3, 133 

Cathedral, Marseilles 147 

Cathedral of Notre Dame, Paris 146 
Cathedral of St. Paul, London 

77, 160 

Cathedral of York 64 

Chamber of Commerce, New 

York 193 

Chateau d' Azay-le-Rideau .... 90 

Chateau de Chaumont 207 

Chateau de Chenonceaux 206 

Chateau de Josselyn 76 

Chateau de Villebon Si 



Chaumont, Chateau de ^07 

Chenonceaux, Chateau de 206 

rbourg Apartment House, 

New York 67 

Church, All Souls 40 

Church of St. Ambroise, Paris 224 

Church of St. Nicholas, Pots- 
dam 148 

Church. St. Peter's 122 

dar forms in plan and 
elevation 41 

Classification of all buildings as 

to primary massing 85 

■ Li m of arts 13 

ifications of horizontal 
divisions 125 

Colonnade, continuity of num- 
ber and line 105 

Comedy Theatre, Berlin 66 

m and criticism 212 

parison, Vienna Rathhaus 
with gateway 213 

Compositions of details analo- 
gous to parts of first order. . no 

Composition of paintings and 

buildings compared 81 

ructed decoration 9 

Continuity contrasted with in- 
dividuality 22 

Continuity of effect of four or 

ire details 103 

Continuity in horizontal divi- 
sion 132 

Continuity, horizontal lines re- 
quired for 27 

Continuity in number 

ciated with horizontal lines. 104 

Continuity in numbers exceed- 
ing three 96 

Continuity in primary masses 
impracticable 97 

Continuous details, summary 
of rules for use 107 

Continuous ornament in two 
dimensions 106 


( ' »ntradictory subordination to 

be avoided 52 

Contrast 170 

I ontrast between details 176 

Contrast between masses and 

their links and appendages . . 173 
Contrast between primary and 

secondary masses 175 

Contrast of form 171 

Contrast, general rule for 177 

Contrast, various kinds of 170 

Cornice, ill effect of broken. . . 106 

Cornice used as link 68 

Criticism, Palace of Ludwig 

Victor, Vienna 221 

Criticism, RingTheatre, Vienna 215 

Criticism, Vienna Rathhaus.. . 212 

Decoration, constructed 9 

Definitions of architecture .... 8 
Details, analogy between pri- 
mary masses and 103 

Details, compositions analogous 

to parts of first order no 

Details, four or more produce 

effect of continuity 103 

Details, proportion of 151 

Details, single 99 

Details, summary of rules for 

continuous 107 

Details, technical definition ... 99 
Details, technical use of the 

term 62 

Details, vertical 112 

Detroit, Mich., house at 71 

Diagram of all possible modes 

of horizontal division 127 

Diagram of classification by 

horizontal divisions 125 

Diagram of classification by 

primary massing 85 

Diagram of development of 

motives 203 

Division, analogy between ver- 
tical and horizontal 132 


2 2() 

r \ci. 

Division, continuity in hori- 
zontal 132 

Division into horizontal parts. 115 

Division into three horizontal 
parts 122 

Division into four horizontal 
parts 128 

Division, horizontal, roofs. ... 134 

Domes harmonize best with 
circular forms 40 

Dome, subordinate or domi- 
nant 47 

Dorilton Apartment House, 
New York 68 

Dormers, similarity in their 
forms 39 

Dormitory, Princeton, X.J... 91 

Double horizontal division .... 121 

Double primary masses, asym- 
metrical 70 

Double primary masses, ill 
effect of dissimilarity 69 

Double primary masses, in- 
stances of 64 

Double primary masses, plans 
leading to 179 

Double primary masses, prac- 
tical designing 184 

Double primary masses, simi- 
larity essential 68 

Double primary masses, some 
connection needed 65 

Double secondary masses 87 

Double secondary masses, 
asymmetrical 91 

Double secondary masses, prac- 
tical designing 185 

Elevation, compositions must 
be worked out in 57 

Farnese Palace 152 

Farnese Palace, tabulated pro- 
portions 151 

Flexibilitv of motives 202 

Flexibility of motives, diagram 
of development 203 

Four horizontal parts 128 

Four primary masses, analogy 
in musical rhythm 80 

Four primary masses, of rare 
occurrence 79 

Four secondary masses 94 

Gardener's cottage, Long Is- 
land, New York 199 

Gate Lodge, North Easton, 
.Mass 192 

Greek enriched mouldings, 
similarity of 46 

Greek forms nonrepresenta- 
tive 15 

Greek temple, proportions of . . 156 

Greek temple, similarity of 
parts of 45 

Height as related to width .... 29 
Heights, subordination marked 

by different 49 

Horizontal division 115 

Horizontal division, continuity 

in 132 

Horizontal division, diagram of 

classification 125 

Horizontal division, diagram of 

all possible modes 127 

Horizontal division into four 

parts 128 

Horizontal divisions, height of 

double 163 

Horizontal divisions, height of 

triple 166 

Horizontal division, methods 

of treatment 126 

Horizontal division, roofs 134 

Horizontal division, single. ... 1 20 
Horizontal divisions, summary 

of rules for heights 169 

Horizontal division into three 

parts 122 



Hon ' division into two 

parts i 21 

Hon ■ 1 1 1 ;l! lines, erroneous use 


Horizontal lines essential to 

the unity of continuity 27 

Hon ontal and vertical divi- 
sion, analogy between 127 

Horizontal <»r vertical lines 

must predominate 120 

Horizontal and vertical treat- 
it 30 

Hdtel de Ville, Lyons 65 

Hotel de Ville, Paris 156 

lintel de Ville, Paris, tabulated 

proportions 155 

Hotel Essex, New York 194 

House at Kansas City, Mis- 
souri 71, 195 

House at Tuxedo, New York. . 92 
House at University Heights, 
New York 87 

Individuality and continuity. . 22 
Individual parts, subordina- 
tion or dominance of 58 

Ionic capital compared with 
analogous primary massing. 103 

Jefferson Market Court House. 29 
Josselyn, Chateau de 76 

Library at Quincy, Mass 89 

Limit of size of secondary 
masses 92 

Link essential for connecting 
primary masses 65 

Links 60 

Links and masses, proportions 
of 146 

Links not required with sec- 
ondary masses 97 

Links sometimes used with 
secondary masses 98 





Links used to connect primary 


Lintel, imitating fiat arch. . . . 
Loches, Porte des Cordeliers . . 
Louis XV style, similarity of 


Louvre, Paris 1 

Lucca, Campanile 133 

Lyons, Bourse 76 

Lyons, Hotel de Ville 65 

Madonna, Sistine 82 

Marco Palmizzano, Nativity. . 83 

Marseilles, The Bourse 121 

Marseilles, Cathedral 147 

Masses and appendages, pro- 
portions of 144 

Masses, double primary, asym- 
metrical 70 

Masses, double primary, con- 
nective part essential 65 

Masses, double primary, ill 

effect of dissimilarity 69 

Masses, double primary, simi- 
larity essential 68 

Masses, four primary, parallel 

in musical rhythm 80 

Masses, four primary, rare 

occurrence 79 

Masses, four secondary 94 

Masses and links, proportions 

of...- 145 

Masses, primary, instance of 

double 64 

Masses, primary, instances of 

single 63 

Masses, primary, use of the 

term 59 

Masses, secondary 86 

Masses, secondary, asymmet- 
rically placed 88 

Masses, secondary, limit of size 92 
Masses, secondary, use of the 

term 59 

Masses, summary of rules 78 




Masses, three secondary 91 

Masses, triple primary 73 

Masses, triple primary, analo- 
gies in musical rhythm 76 

Masses, triple primary, central 

must be the largest 74 

Masses, triple primary, sum- 
mary of rules 78 

Masses, two asymmetrical pri- 
mary 19S 

Masses, two asymmetrical 

secondary 201 

Materials, effect upon design. . 9 

Medici, Villa , 145 

Mont St. Michel 28 

Morgan Art Gallery 51 

Morris Heights School, New 

York 75 

Mouldings, apparently redun- 
dant 130 

Mouldings, undercut 118 

Mouldings used to mark hori- 
zontal divisions 115 

Music compared with archi- 
tecture 15 

Nativity by Marco Palmizzano 83 
North Easton, Mass., gate 

lodge 192 

Notre Dame, Paris 146 

Oblique point of view 57 

Office building, St. James, New 

York 124 

Office building, St. Paul, Minn. 72 
Office building, St. Paul, New 

York 133 

Overhang, used to mark hori- 
zontal divisions 116 

Paestum, temple 162 

Pagoda, Tokio 133 

Paintings, comparison with 

composition of buildings. ... 81 

Palace of Ludwig Victor, Vi- 
enna. . 221 

Palazzo Vecchio 149 

Palais Epstein, Vienna 129 

Palais de Justice, Antwerp. . . 77 
Paris, Cathedral of Notre Dame 1 46 

Paris, Hotel de Ville 156 

Paris, The Louvre. 121 

Parliament Buildings, West- 
minster 73 

Philae, Temple at 64 

Piazza del Popolo, Rome 67 

Pilasters, used to mark vertical 

lines 28 

Pisa, Cathedral 166 

Pisa, Cathedral, height of hori- 
zontal divisions 167 

Pitti Palace 109 

Planning 178 

Plan for double masses 179 

Plan for triple masses 1 80 

Polka dot, exhibits continuity 

in two dimensions 106 

Porte des Cordeliers, Loches... 205 
Portico, typical continuous 

form 27 

Practical designing 1S3 

Practical designing of double 

primary masses 184 

Practical designing of double 

secondary masses 1S5 

Practical designing of single 

secondary mass 187 

Primary masses 59 

Primary masses, analogy be- 
tween details and 103 

Primary masses, buildings 

classified according to S5 

Primary masses, connecting 

link essential 65 

Primary masses connected by 

linking parts 60 

Primary masses, double 64 

Primary masses, double, asym- 
metrical 7° 

2 }2 


Primary masses, double, prac- 
tical designing 184 

Primary masses, double sum- 
mary of rules 7 s 

Primary masses, four, analogy 

in rhythm 80 

Primary masses, four, rare oc- 
currence 79 

Primary masses, more than 

three impracticable 97 

Primary masses, single 63 

Primary masses, triple 73 

Primary masses, triple, central 

must be the largest 74 

Primary masses, triple, sum- 
mary of rules 78 

Primary masses, two asym- 
metrical 198 

Princeton, N. J., dormitory.. . 91 

Proportion 137 

Proportion, architectural same 

as geometrical 138 

Proportion of details 150 

Proportion, example of defec- 
tive 200 

Proportion, examples of vary- 
ing 210 

Proportion, instances of lack of 147 
Proportion of masses and ap- 
pendages 144 

Proportion of masses and links. 145 
Proportion, method of diag- 
onals 140 

1'p ipi »rtion 1 »f roofs 158 

Proportion of secondary 

masses 149 

Proportion of St. Paul's, Lon- 


I'm lit zer I louse 37 

Pure art compared with rep- 

entative 12 

Pyramid, typical individual 

form 24 

Pyramidal arrangement of 

parts 28 

Quincy, Mass., library at. 

Raphael, Sistine Madonna.. . . 82 

Raphael, Transfiguration 84 

Rathhaus, Vienna 212 

Relation of architecture to con- 
struction 9 

Representative art and pure 

art 12 

Rhythm, analogy in quad- 
ruple massing 80 

Rhythm, analogies in triple 

massing 76 

Rhythm compared in various 

arts 17 

Ring Theatre, Vienna, criti- 
cism of 215 

Roofs, considered as horizontal 

divisions 134 

Roofs, individual or continu- 
ous views 34 

Roofs, mansard 135 

Roofs, proportion of 158 

Roofs, pyramidal and hori- 
zontal 135 

Roofs, similarity of treatment 
essential ^^ 

School, Morris Heights, New 
York 75 

School, St. Stephen's, New 
York 1 00 

Secondary masses 59, 86 

Secondary masses, double 87 

Secondary masses, double 
asymmetrical 91, 200 

Secondary masses, double, 
practical designing 185 

Secondary masses, four 94 

Secondary masses, limit of 
size 92 

Secondary masses, proportions 
of 149 

Secondary masses, single asym- 
metrically placed 88 



Secondary mass, single, prac- 
tical designing i s - 

Secondary masses, triple 91 

Similarity both in plan and 

elevation 41 

Similarity of circular forms ... 40 
Similarity comparable with mu- 
sical unison 37 

Similarity contrasted with dis- 
similarity 43 

Similarity of dormers 39 

Similarity, essential to unity. . 35 
Similarity as evinced in classi- 
cal and Gothic styles 44 

Similarity, painful effect when 

lacking 43 

Similarity of parts of Greek 

temple 45 

Similarity of parts in Louis X V 

style 46 

Similarity of parts the main 

characteristic of style 42 

Similarity of window openings . 3 6 

Single details 99 

Single primary masses, in- 
stances of 63 

Single secondary masses 88 

Sistine Madonna 82 

Standard of taste 3 

St. James office building, New 

York 124 

St. Mark's Campanile 3, 133 

Stoedel Art Gallery, Frankfort- 

on-the-Main 77 

St. Paul's Cathedral, London. . 77 
St. Paul's, London, proportion 

of 160 

St. Paul, office building at . . . . 72 
St. Paul office building, New 

York 133 

St. Peter's Church, Rome 122 

Strozzi Palace 123 

St. Stephen's School, New York 100 
Style, characterized by simi- 
larity of parts 42 

Subordination, contradictory.. 52 

Sub. irdination by difference in 
height 1 , 1 

Subordination bydifference in 
projection 53 

Subordination by difference in 
width ^o 

Subordination, different meth- 
ods of securing (.8 

Subordination of parts essen- 
tial for unity 47 

Subordination of portico 48 

Subordination to a single domi- 
nant part 56 

Synagogue, Berlin 149 

Taj Mahal 50 

Temple at Paestum 1 02 

Temple at Philae 64 

Three-part horizontal division . 122 

Three primary masses 73 

Three primary masses, central 

must be the largest 74 

Three primary masses, rhythm 

analogy 76 

Three primary masses, sum- 
mar)' of rules 78 

Three secondary masses 91 

Tower, typical individual form. 25 
Townsend House, Washing- 
ton, D. C 144 

Transfiguration, by Raphael. . 84 
Triglyphs forming successi< >n 

of vertical lines 119 

Triglyphs, omitted in later 

orders 5 

Two asymmetrical append- 
ages 196 

Two asymmetrical primary 

masses 70 

T\v< 1 asymmetrical secondary 

masses 205 

Two-part horizontal division. . 121 
Two primary masses, connect- 
ing link essential 65 


Two primary masses, ill effect 

of dissimilarity 69 

Two primary masses, instano 


primary masses, similarity 

ntial 6S 

primary masses, sum- 
mary of rules 78 

Undercut mouldings 118 

Unequal primary masses 70 

Unity, the essential quality in 

art 19 

Unity, individual and continu- 
ous 24 

Unity obtained by similarity 

of parts .' '. 35 

Unity obtained by subordina- 
tion of parts 47 

Variations of proportion 210 

Vendramini Palace 36 

Vertical details 112 

Vertical and horizontal divi- 
sion, analogy between 127 


Vertical and horizontal lines 

must predominate 120 

Vertical or horizontal treat- 
ment 30 

Vertical lines, marked by pilas- 
ters 28 

Vienna, Academy of Fine Arts. 128 
Vienna, Palace of Ludwig 

Victor 221 

Vienna, Palais Epstein 129 

Vienna, Rathhaus 212 

Vienna, Ring Theatre 215 

Villa Albani, arcade 42 

Villa Borghese 145 

Villa Medici -. 145 

VilleDo'n, Chateau de 81 

Westminster Parliament Build- 
ings 73 

Whitehall, Banqueting Hall. . . 120 

Width as related to height .... 29 
Window openings, similarity 

of 36 

York Cathedral 64 


University of California 


405 Hilgard Avenue, Los Angeles, CA 90024-1388 

Return this material to the library 

from which it was borrowed. 

<JL APR 15 1996 

PSD 23G 


" CAt/Fl 


owmin^ 'owmm 

% *^ / 1 

c^EliNIVER^ v lOSAVCfl/| 


3? ~~ r w * 



Wtfl'NIVERS/^. ^aOS-ANGElfj> 








NA 2760 R563a 

L 005 861 882 8 


y 0Aavn8ii-# 





^^^-> a- 

<TJl33WS0^ N 






<TiirjNVSO^ N 


















.\VlE-UNIVER% il)SANCElfj>